i

TWO ESSAYS ON U.S. GRAIN EXPORT PROMOTION

A Thesis

Presented to the Faculty of the Graduate School

of Cornell University

In Partial Fulfillment of the Requirements for the Degree of

Master of Science

by

Pimbucha Rusmevichientong

January 2008

ii

© 2008 Pimbucha Rusmevichientong

iii

ABSTRACT

More than one-third of agricultural production in the United States is exported.

U.S. rice production has been increasing every year for decades in response to

growing world rice trade and a relatively tight supply in other exporting countries. The

U.S. competes with other major rice exporters such as Thailand, Vietnam, China, India

and Pakistan.

The United States Department of Agriculture (USDA) has administered two

export promotion programs to encourage the development, maintenance and

expansion of international market of U.S. rice: The Foreign Market Development

Program (FMD) in 1984 and the Market Access Program (MAP) in 1986. USA Rice

Federation (USARF) and USA Rice Producer Association (USRPA) are the only two

private money recipients. Those two public-private cost-share programs generate over

$4 million annually to promote all varieties of U.S. rice.

This study investigates the economic impacts of U.S. rice export promotion

program on U.S. rice export demand. The overall effectiveness of the export

promotion program is evaluated in terms of benefits relative to costs. The optimality

of export promotion expenditure levels is computed by computing a marginal benefit-

cost ratio. The results indicate that U.S. rice export promotion has been effective in

enhancing foreign market demand for U.S. rice and the U.S. is under-investing on rice

export promotion.

However, these programs remain highly controversial with opponents claiming

they amount to “corporate welfare”. Government intervention in funding export

promotion can be rationalized in an economic standpoint if market failure exists. One

i

type of market failure, “positive externalities or halo effects” produced by export

promotion of one commodity increasing the export demand for other U.S.

commodities are not captured in a single equation demand. Also, foreign market

promotion constitutes a type of “international public good or free-riding effect” when

it has positive cross effects that dominate country-specific effects. Failing to take

these cross effects into account tends to generate bias results of the promotion.

A dynamic linear approximation of an Almost Ideal Demand System (AIDS)

model with Seemingly Unrelated Regression (SUR) method is employed to estimate

the U.S. and non-U.S. grain market shares from 1975 to 2005 incorporated with major

grain exports, i.e. rice, wheat and sorghum to examine the issues concerning cross-

promotional effects. The simulation results indicate no halo effects of U.S. grain

demand on rice, wheat, and sorghum export promotion; however, U.S. grain export

promotion presents an anti-halo effect on competing country grain exports. Hence,

there is no international free riding of U.S. grain export promotion.

iii

BIOGRAPHICAL SKETCH

Pimbucha Rusmevichientong was born in Bangkok, Thailand on January 27th

1983 and received a Bachelor of Economics degree for her studies in the field of public

finances in May 2005 from the Department of Economics at Chulalongkorn University,

Thailand. In 2006, she received the Royal Thai Government Scholarship to pursue a

course of studying leading to Master’s degree in Applied Economics and Management

at Cornell University. After graduation in 2007, she will return to her post in

Cooperative Promotion Department, Ministry of Agriculture and Cooperatives where

she is legally obliged to work.

Initial exposure to agricultural economics at Cornell University was

challenging; however, it revealed a broad range of enthralling exploration in the area to

her. This led her to get involved in many great academic opportunities to improve and

develop her knowledge and research skill. Living in this fantastic gorge area one year

and a half has given invaluable experiences to her not only in academia but also in

learning how to live in multicultural society. Studying at Cornell and working with her

advisor inspire her to further study in PhD in the future.

iv

To my family and friends

v

ACKNOWLEDGMENTS

I am greatly indebted to the Royal Thai Government for offering a great

opportunity to me through financial supports to study at Cornell University.

I gratefully acknowledge my advisor, Professor Harry M. Kaiser, for his

insightful and generous guidance. His academic comments, scrupulous editing and

great academic opportunity offerings, i.e., paper presentation in many conferences

helped polish my analytic skill, refine my thesis and complete my Master’s degree

early. In addition, I am thankful to my minor advisor, Professor David R. Just for his

expert advice on econometrics behind the models. I also appreciate the patience of

Yuqing Zheng, Professor Timothy J. Richards, Professor Henry W. Kinnucan, and Xie

Jinghua, who read drafts and gave invaluable feedbacks. Their thoughtful and specific

comments helped me make revisions to improve my papers. Also I would like to thank

you Anita Vogel for her helpful administrative supports.

I am very thankful to many professors in Applied Economics and Management

Department with whom I have taken courses to give knowledge and motivation on my

thesis.

I also wish to acknowledge the importance of room 37 in Warren Hall, which

made my experiences as a graduate student less lonely, and created a venue for fun and

academic support. My thanks to fellow graduate students for their feedbacks and

supports on my research, Apurba Shee, Brock Williams, Esther Ashburn, Fahran Robb,

Ghazi Alam, Kira Villa, Kiran Gajwani, Leslie Verteramo, Margaret Ang, Matthew

LeRoux, Michel Norton, Xinhua Yu, and other students I have not mentioned who

sometimes stop by at Room 37 to find cozy environment and enjoyment beyond term

papers and assignments.

Last but not least, I would like to thank my family for their endless love,

continuous support and encouragement throughout all my life.

vi

TABLE OF CONTENTS

BIOGRAPHICAL SKETCH……………………………….…….…………………..III

DEDICATION……………………………………………….…………………...…..IV

ACKNOWLEDGEMENT…………………………………….………….…………...V

TABLE OF CONTENTS……………………………………….…………….……...VI

LIST OF FIGURES…………………………………………….………………......VIII

LIST OF TABLES…………………...………………………….……………………IX

CHAPTER 1: INTRODUCTION ……..……………………….…...…………….…...1

1.1 U.S. RICE EXPORT PROMOTION PROGRAMS…….………………….......2

1.2 PURPOSE……………………………………………….……………………...3

1.3 OBJECTIVES…………………………………………….……………….........4

1.4 OUTLINE OF THE THESIS…………………….……….………………..…...5

CHAPTER 2: LITERATURE REVIEW……..………...………..…………………….6

CHAPTER 3: MEASURING THE EFFECITIVENESS OF U.S. RICE EXPORT

PROMOTION PROGRAM..……………...…………………….…………….……...15

3.1 INTRODUCTION……………..……………….…………………….15

3.2 PREVIOUS RESEARCH ...………………………………................18

3.3 ECONOMETRIC MODEL…………………….…………………….21

3.4 ECONOMETRIC RESULTS………………….……………......……23

3.5 SIMULATION ANALYSIS………………….....……………....…...25

3.5.1 Average Benefit-Cost Ratio………………………..…………...............27

3.5.2 Marginal Benefit-Cost Ratio…………………………..............……….31

3.6 CONCLUSIONS……………………….…….………….…………...34

REFERENCES………………………………......………………..………......36

vii

CHAPTER 4: THE HALO EFFECT OF U.S. GRAIN EXPORT PROMOTION......40

4.1 INTRODUCTION…………………………………………………..….…......40

4.2 THE LINEAR APPROXIMATION ALMOST IDEAL DEMAND SYSTEM

(LA/AIDS)……..………...…………………………..……….………………….…...43

4.3 DATA……………………………..…………………..…………...………….47

4.4 ECONOMETRIC RESULTS………………..………..……………….……...51

4.5 SIMULATIONS……………..………………………………...………..…….60

4.6 SUMMARY AND CONCLUSIONS……………..………………..…………63

REFERENCES………………..…………………………….……..…………….........65

CHAPTER 5: SUMMARY AND CONCLUSIONS. ……….……………………….67

viii

LIST OF FIGURES

Figure 3.1: Simulated U.S. Rice Exports with and without Export Promotions ……26

ix

LIST OF TABLES

Table 2.1: Results from economic impact studies on U.S. export promotion using a

single equation estimation…………………….…….…………….……....12

Table 2.2: Results from economic impact studies on U.S. export promotion using an

integrated framework…………………………………..…….……..….….13

Table 3.1: Average annual world market impacts and benefit-cost ratios due to U.S.

rice export promotion, 1984-2005……………………..……...……….….30

Table 3.2: Lower-bound 95 percent confidence interval for benefit-cost ratios due to

U.S. rice export promotion, 1984-2005……………….……………..….…….31

Table 3.3: Marginal benefit-cost ratios due to U.S. rice export promotion,

1984-2005……………………………………...……………………….…......31

Table 3.4: Key results from economic impact studies on U.S. export promotion….....33

Table 4.1: An average export volumes and values of major exporters, rest of the world

and world, 1975-2005……………………………………………………........48

Table 4.2: Variable definitions and summary statistics, 1975-2005………………..…50

Table 4.3: Price coefficients of LA/AIDS for U.S. grains, 1975-2005……………......53

Table 4.4: Promotion coefficients of LA/AIDS for U.S. grains, 1975-200…………...53

Table4.5: Intercept, expenditure, trend, adjusted R2 and Durbin – h value of LA/AIDS

for U.S. grains, 1975-2005…………………………....………………........54

Table 4.6: Estimated short-run compensated price elasticities of U.S. and non-U.S.

rice, wheat and sorghum, 1975 -2005…………………………..…………56

Table 4.7: Short-run compensated promotion and expenditure elasticities of U.S. and

non-U.S. rice, wheat and sorghum, 1975 -2005…………………..……....56

Table4.8: Long- run compensated price, export promotion and expenditure elasticities

of U.S. and non-U.S. rice, wheat and sorghum, 1975 -2005….……………57

x

Table 4.9: Long- run promotion and expenditure elasticities of U.S. and non-U.S. rice,

wheat and sorghum, 1975 -2005……………………...……………............57

Table 4.10: An average direct and total benefit-cost ratios for U.S. rice, wheat, and

sorghum export promotion (1990-2005)……………………….………...62

1

CHAPTER 1

INTRODUCTION

Rice production in the United States has been increasing every year for decades

in response to growing world rice trade and relatively tight supplies in other exporting

and importing countries. From the late 1960’s through 1980, the U.S. was the largest

world rice exporter, with Thailand occasionally out-shipping the United States;

however, since 1981, Thailand has been the leading rice exporter, largely due to

expanding rice crop area. By the mid-1990’s, Vietnam replaced the U.S. as the second

largest rice exporter after it had recovered from the war (1957-1975) and political

upheavals. China soon after became the third largest exporter due to declining per

capita domestic consumption, and India emerged as a major rice exporter, ranking

forth or fifth each year.

The global rice market is segregated by type and quality, with little substitution

among buyers (Childs and Burdett, 2000). In fact, tastes and preferences are so strong

that prices for various types of rice can possibly move in opposite directions. There is

little substitution in production among the various types of rice either, as soil and

climate often dictate the type of rice that can economically be grown in any particular

area. As a result, global rice prices are typically more volatile than prices for other

grains.

Traded rice can be categorized by physical type (long grain, medium and short

grain), quality (high and low quality by kernel brokenness), and degree of milling

(paddy or rough rice, brown rice, milled rice and parboiled rice). Milled rice accounts

for the bulk of global rice trade, with brown rice ranking second. Very little rough rice

is traded, as most countries prefer to either capture the value added from the additional

processing. However, some countries e.g. Mexico, prefer to import mostly rough rice

2

to support their domestic milling industries. The United States is the major rough rice

exporter. Rough, brown and milled rice can be exported as parboiled1 which is

typically sold at a premium to non-parboiled. The major exporters of parboiled rice are

Thailand, India, and the United States. The Middle East, Western Europe, and South

Africa are the main markets for parboiled rice.

1.1 U.S. Rice Export promotion programs

To facilitate U.S. rice exports, three main types of government programs assist

U.S. rice exports expansion. First, P.L. 480, Section 416(b) and Food for Progress etc.

are the programs the United States sells rice on concessional credit terms and donates

rice to needy countries bilaterally or through the World Food Program. Second, the

U.S. Department of Agriculture (USDA) provides export credit guarantees for

commercial financing of U.S. agricultural export, i.e. GSM- 102, GSM-103, and

Supplier Credit Guarantee Program. Third, the creation, expansion, and maintenance of

foreign markets for U.S. rice, which are the main focus of the paper, are funded by the

USDA Market Development Program – Market Access Program (MAP) and Foreign

Market Development Program (FMD). Several other programs such as the Emerging

Market Program, the Qualities Samples Pilot Program, the Cochran Fellowship

Program, and Section 108 also provide assistance to U.S. rice exports and/or market

share expansion.

Market development programs – MAP (Market Access Program) and FMD

(Foreign Market Development) are partnership programs between FAS/USDA and

numerous non-profit private sector commodity and regional associations. Created in

1955, FMD embodies the agency’s primary goal to develop, maintain and expand

1 Parboiling is a process whereby rough rice is soaked in water and steamed under intense pressure. Parboiling makes the rice less likely to break during milling and pushes nutrients from the bran layer into the kernel.

3

long-term export markets for U.S. agricultural products and is generally used for bulk

and generic products. One-half of U.S. government funding under FMD was spent on

feed grains, wheat and soybeans. The remaining funds went to forest products, meat,

rice and poultry exporters. The MAP program was originally started in 1985 as the

Targeted Export Assistance Program (TEA) to offset the adverse impact of unfair trade

practices of other countries on exporters of American agricultural products. Like FMD,

the MAP program also helps U.S. producers, exporters, private companies and other

trade organizations finance promotional activities for U.S. agricultural products.

Unlike FMD, MAP is intended to be used for shorter- term, consumer oriented

promotions and is primarily used for high value and processed products as well as

branded promotions. USA Rice Federation (USARF) and USA Rice Producer

Association (USRPA) are the only two money recipients for rice export promotion.

These two public-private cost-share programs generate over $4 million annually to

promote US rice in all varieties

1.2 Purpose

The purpose of this thesis is to investigate the economic impacts of U.S. rice

and grain export promotions on U.S. rice export demand and U.S. grain export market

shares. Export markets are important to U.S. agriculture accounting for 20% – 30% of

total annual agricultural production over the past twenty years (Benson and Marchant

(1996)). Export promotion is one of the major factors influencing US agricultural

exports; however, questions remain concerning its economic impacts.

Numerous studies have been conducted on various commodities and individual

countries, but few have focused on grain markets. This study is the first to estimate

U.S. grain export promotion programs impact, i.e. rice and other related grains, on the

U.S. grain export markets and evaluate a benefit-cost ratio as a bottom-line measure of

4

promotion effectiveness of the program. To our knowledge, no study has estimated the

effect of U.S. grain export promotion programs on grain exports as opposed to exports

in targeted countries. This thesis provides empirical evidence on the impact of U.S.

grain promotion programs and their net benefits.

1.3 Objectives

The main objective of the thesis is to analyze the economic impacts of U.S.

grain export promotion programs (rice, wheat and sorghum). The specific objectives

are:

1. to estimate the responsiveness of U.S. rice export demand with respect to

U.S. rice export promotion.

2. to evaluate the overall effectiveness of the U.S. rice export promotion

program as to whether it has been profitable to U.S. rice growers by

evaluating the benefits relative to the costs.

3. to determine the optimality of export promotion expenditure levels regarding

whether the U.S. has under- or over-invested in U.S. rice export promotion

by computing the marginal benefit-cost ratio for the program.

4. to examine whether there is a “halo effect” of U.S. grain (rice, wheat, and

sorghum) export promotion, i.e., test the cross-effects of one grain’s

promotion (e.g., U.S. rice promotion) on another grain’s (e.g., U.S. wheat)

market share.

5. to investigate whether U.S. grain export promotion has public goods

characteristics, i.e., test the cross-effects of U.S. grain export promotion on

non-U.S. grain market share (other countries).

5

1.4 Outline of the thesis

The thesis consists of two essays on U.S. grain export promotion. It is organized

as follows: Chapter 2 consists of a literature review of previous research on U.S. export

promotion programs for various commodities; Chapter 3 presents the first essay:

Measuring the Effectiveness of U.S. Rice Export Promotion Program. This paper

provides an econometric investigation on the responsiveness of U.S. rice export

demand with respect to U.S. rice export promotion program and the effectiveness of

the program; Chapter 4 presents the second essay: The Halo Effects of U.S. Grain

Export Promotion. The study develops an integrated framework to examine the own-

and cross-effects of U.S. grain export promotions for U.S. and non-U.S. rice and

related grains, (wheat and sorghum). Finally, Chapter 5 provides the conclusion for the

two essays.

6

CHAPTER 2

LITERATURE REVIEW

There have been numerous studies that examine the impact of the promotion

programs in various commodities and individual country specific in term of promotion

elasticity, generally conducted by econometric estimation. Most studies employed a

single-equation with (per capita) import or export demand as a dependent variable and

demand determinants as an explanatory variables.

Kaiser, Liu, and Consignado (2005) studied the effectiveness of the Raisin

Administrative Committee (RAC)’s export promotion programs between 1965-1998 to

Japan and the United Kingdom. A single equation import demand for California raisins

in both countries was estimated to determine the responsiveness with respect to the

RAC’s export promotion, which included both government and private contribution.

The estimated promotion elasticity of Japan and the U.K. was 0.029 and 0.133

respectively. Average and marginal benefit-cost ratios were estimated as a bottom line

of effectiveness measurement. The average gross benefit-cost ratio was 5.13 for Japan

and 15.29 for the U.K. The marginal gross benefit-cost ratio was 0.42 for Japan and

3.19 for the U.K.

Lanclos, Devodoss and Guenthner (1997) chose representative importing

countries for U.S. frozen potatoes from four tiers based on 1983 import volume: Japan

(tier1), Mexico (tier2), the Philippines (tier3) and Thailand (tier4). Geweke’s Bayesian

estimation procedure was used to estimate each single equation import demand

function for the four countries from 1978-93, allowing for the incorporation of prior

information into the coefficient estimation. The results indicated that U.S. Potato

Board Advertising had very small impacts in each country with elasticities of 0.03 in

Japan, 0.04 in Mexico, 0.06 in the Philippines, and 0.08 in Thailand. On the other

7

hand, Third Party Program advertising had much larger impacts with advertising

elasticities of 0.03 in Japan, 0.53 in the Philippines, and 0.87 in Thailand. The marginal

benefit-cost ratio for the Potato Board and Third Party advertising was 1.3 and 1.29 in

Japan, 1.42 and 11.77 in the Philippines, and 1.51 and 16.36 in Thailand. For Mexico,

a marginal benefit-cost ratio for the Potato Board was 1.13.

Fuller, Bello and Capps (1992) estimated a single equation per capita demand

for U.S. fresh grapefruit with quarterly data from 1969-88 for Japan, Canada, France,

and the Netherlands. They regressed per capita imports for each country on the U.S.

fresh grapefruit price F.O.B, the exchange rate between the importing country’s

currency and the U.S dollar, per capita GDP in each importing country, the price of

grapefruit substitutes (banana) in each importing country, a trend term, seasonal

dummy variables, several policy variables reflecting tariff and quota trade barriers, and

U.S. export promotion expenditures to each country. The promotion elasticity was

estimated in France (0.234), the Netherlands (0.153) and Japan (0.109). The marginal

benefit-cost ratios for the three countries were 6.65 for the Netherlands, 5.02 for Japan,

and 4.13 for France.

Halliburton and Henneberry (H-H) (1995) examined the effectiveness of the

FMD and MAP programs from 1986-92 for almonds in Japan, South Korea, Hong

Kong, Taiwan, and Singapore. Import price of almond substitutes and compliments

(cashews, sugar, and cocoa butter) were included in the model. H-H estimated a single

equation with three different functional forms: logarithmic, linear, and exponential

forms. The results in terms of export promotion effectiveness were mixed. Using the

promotion elasticities from the linear model, H-H computed marginal benefit-cost

ratios of 4.95 for Japan, 5.94 for Taiwan, and 3.69 for Hong Kong.

Le, Kaiser and Tomek (1998) estimated the economic impact of FAS sponsored

red meat export promotion on a single equation import demand in Hong Kong, South

8

Korea, Singapore, and Taiwan. Panel data (pooled time-series and cross-section data)

were used to estimate an import demand for red meat which included export promotion

expenditures for FMD and TEA combined as one of the explanatory variables. There

was a positive and significant impact of red meat export promotion on South Korea

(0.598), but not in other countries. The gross marginal benefit costs ratio calculated for

four countries ranged 15.62 and 47.32

Unlike many studies that estimated single equation import demand equations for

countries, Rosson, Hammig and Jones (1986)) examined a U.S. apple export demand

equation which include export promotion expenditure from FAS program as well as

U.S. cooperator and foreign agent funds. However, this study did not include income,

competing promotion or other demand factors due to the limited number of

observation. The results indicated that over the period 1974-1981, the export

promotion elasticity for apples was 0.51, and the marginal benefit-cost ratio was 60.

Onunkwo and Epperson (2000) also investigated the impacts of FAS promotion

programs on U.S. pecan exports to Asia and the EU. While most single equation

studies have included export promotion expenditures for the specific commodity being

investigated, this study included, as separate variables, promotion expenditures for

pecans, walnuts, and almonds. Thus, any indirect or spill-in effect of almond and

walnut promotion on U.S. pecan exports could be calculated. The export promotion

elasticities were statistically significant and positive in Asia (0.98) and the EU (0.06).

In Asia, U.S. almond and walnut promotion were actually larger than the direct effect

of pecan promotion on pecan export. U.S. walnut promotion had a negative impact

(cross-promotion elasticity of -1.64) while U.S. almond promotion helped U.S. pecan

export with cross-promotion elasticity of 1.38. In the EU, the opposite results were

found with walnut promotion helping (0.48) and almond promotion harming (-0.14)

U.S. pecan exports. The Asian export promotion elasticity was over 16 times larger

9

than the EU; this study reported a higher gross marginal benefit-cost ratio for the EU

(6.75) compared with Asia (6.45)

Armah and Epperson (1997) estimated a single equation export demand for U.S.

orange juice using cross sectional country (Japan and four European countries: France,

Germany, the Netherlands, and the United Kingdom) data pooled with time series over

the period 1984-92. U.S. export promotion gave the largest impact in the Netherlands

with an estimated promotion elasticity of 0.302. Germany was 0.044; while Japan, the

U.K. and France all had elasticities of 0.014. Marginal benefit cost ratios (MBCR)

were calculated for each country: the Netherlands (51.92), Germany (37.10), the U.K.

(7.64), France (7.44) and Japan (5.61).

Alston et al. (1997) estimated the effectiveness of U.S. export promotion of

California table grapes to Asian countries with annual time series data from 1976-94.

An aggregate export demand model was estimated and focused on eight importing

countries: Hong Kong, Taiwan, Malaysia, Singapore, the Philippines, Indonesia,

Thailand, and South Korea. Four import demand models for Asian countries were

estimated using monthly data from 1976-94 for Hong Kong,, Malaysia, Singapore and

Taiwan. Unlike the majority of other individual commodity studies above, Alston

incorporated a supply response into the simulation, and therefore price was allowed to

vary with any promotion induced demand increase in the simulation. The export

demand model was combined with an assumed excess supply elasticity which was

defined in constant elasticity form and equated with predicted demand quantities rather

than econometrically estimating an excess supply function. This study specified a

constant elasticity synthetic export supply function and used four different assumed

own price elasticity of supply (1, 2, 5, and 10) to simulate alternative benefit-cost

ratios. Given an own price elasticity of supply of 5, the average benefit-cost ratio

(present value with 3 percent compounding) for U.S. table grape export promotion was

10

9.1.The marginal benefit-cost ratio was computed to be 4.3 indicating under spending

on export promotion. Benefit-cost ratios were not calculated.

The single equation is the most popular means of evaluation in this area due to

simplicity and data availability i.e. competing export promotion and private industry

promotion expenditures results in only partial measurement with necessarily important

variables; however, it is not comprehensive to investigate the spillover and spill-in

(halo) effect of the promotions (see the summarized results of single equation

estimation in Table 1). Not many studies have used demand systems as an integrated

framework such as the Rotterdam model and the Almost Ideal Demand Systems

(AIDS).

Richards and Paterson (1997) used a dynamic dual model of U.S. export supply

for horticulture commodities: apples, almonds, wine and grapes based with time-series

data from 1984-95 pooled with country-level data for Mexico, Sweden, Norway,

Finland, the United Kingdom, France, Germany, Malaysia, Singapore, the Philippines,

and Japan. The direct effects of U.S. export promotion in term of elasticity for all four

commodities were positive and statistically significant: apples (5.011 in short run,

23.949 in long-run), almonds (0.052 in short-run, 0.387 in long-run), wine (2.096 in

short-run, 2.736 in long-run) and grapes (0.467 in short-run, 0.778 in long-run). Most

cross effects were positive and significant, indicating a halo effect among these

commodities. For some commodities, cross effects were as large as or larger than

direct effects, i.e. the elasticity of almond promotion and graph promotion on apples

was 1.028 in short-run, 3.775 in long-run, and 0.647 in short-run, -4.258 in long-run,

respectively.

Richard, Ispelen, and Kagan (R-I-K) (1997) developed a two-stage linear

Expenditure System (LES)/Almost Ideal Demand System (AIDS) to estimate the

effectiveness of apple export promotion in increasing U.S. market share and total apple

11

import demand for Singapore and the United Kingdom markets. Other fruit products

were included in the system, i.e., banana, orange, grapes imports and other domestic

goods. The feature of the demand system model provided the ability to address the

important issue of spillover effects of export promotion and be consistent with the

adding – up, homogeneity, and symmetry restriction of consumer demand theory.

Estimated U.S. apple promotion elasticities on U.S. market share in Singapore and the

U.K. were significantly positive of 0.055 and 0.016 respectively. Cross-effects

indicated that U.S. apple export promotion helped New Zealand (0.079), Chile (0.601),

France (0.186) and Rest of the word (0.048) market share in Singapore, indicating

significantly positive elasticities in those countries. This caused the U.S. to lose market

share to other countries that are able to free ride. U.S. apple export promotion also

provided a halo effect to other countries’ market shares in the U.K. i.e. Chile (0.045),

South Africa (0.034), France (0.079); however, it hurt New Zealand market share in

the U.K. (-0.025). Taking both the aggregate and share effects into account, the total

promotion elasticity for U.S. apples in Singapore was modestly positive comparing

with the U.K. R-I-K suggested that promotion is more effective the less elastic is

demand. Empirically, the effects of export promotion by one country have larger

spillover effects the more inelastic is demand for a rival’s product.

12

* means statistically significant at conventional significance levels, i.e. at least the 10% level.

Table 2.1: Results from economic impact studies on U.S. export promotion using a single equation estimation.

California

raisins US frozen potatoes

US fresh grapefruit

US almonds US red meat US apples US pecans US orange juice CA table grapes

Study

Kaiser, Liu, and Consignado (2005)

Lanclos, Devodoss, and Guenthner (1997)

Fuller, Bello, and Capps (1992)

Halliburton and Henneberry (1995)

Le, Kaiser, and Tomek (1998)

Rosson, Hammig, and Jones (1986)

Onunkwo and Epperson (2000)

Armah and Epperson (1997)

Alston et al. (1997)

US export promotion in:

Japan and UK Japan, Mexico, Philippines, Thailand

Japan, Canada, France, and Netherlands

Japan, Taiwan, Hong Kong, Singapore, South Korea

S. Korea, Taiwan, Hong Kong, Singapore

All countries US has programs in

Asia and EU France, UK, Germany, Japan Netherlands

Asian countries

Period 1965-98 1978-93 1969-88 quarterly

1986-92 panel data

1984-94 panel data

1972-81 1986-96 panel data

1984-92 panel data

1976-94

Model Import demand Import demand Import demand Import demand Import demand Export demand Export demand Export demand Export demand

Estimated promotion elasticities

Japan=0.029* UK=0.133*

Third Party: Japan=0.03* Philippines= 0.53* Thailand=0.87*

Japan=0.109* Netherlands= 0.153* France=0.234*

3 models range from -0.2788 to 0.85

Korea=0.598* HK=-0.019 Taiwan=0.047 Singapore=0.034

Apples=0.51* Asia=0.98* EU=0.06*

France=0.014 Germany=0.044* Japan=0.014 Netherlands= 0.302* UK=0.014*

0.21*

Estimated BCR

Japan: AGBCR=5.13 MGBCR=0.42 UK: AGBCR=15.29 MGBCR=3.19

Third Party MGBCRs: Japan=1.29 Philippines= 11.77 Thailand=16.36

MGBCR: Japan=5.02 Netherlands= 6.65 France=4.13 Canada=no promotions

MGBCRs: Japan=4.95 Taiwan=5.94 HK=3.69

MGBCR: Range from 15.62 to 47.32 for all 4 countries

MGBCR=60

Asia: MGBCR=6.45 EU: MGBCR=6.75

MGBCRs: France=7.44 Germany=37.10 Japan=5.61 Netherlands= 51.92 UK=7.64

ABCR: 4.1-9.4 MBCR: 4.1-4.2

12

13

Table 2.2: Results from economic impact studies on U.S. export promotion using an integrated framework.

US horticultures US apples US non-alcoholic beverage

Study Richards and Patterson (1997) Richard, Ispelen, and Kagan (1997) Xiao, Kinnucan, Kaiser (1999)

US export promotion in:

Mexico, Sweden, Norway, Finland, the United Kingdom, France, Germany, Malaysia, Singapore, the Philippines, and Japan.

a two-stage linear Expenditure System (LES)/Almost Ideal Demand System (AIDS)

milk, juices, soft drinks and coffee and tea

Period 1984-95 pooled with country-level data 1962-93 1970-94

Model A dynamic dual model Import demand, LES/AIDS demand systems

A Rotterdam model

Estimated own promotion elasticities

Apples = SR:5.0*, LR:23.949* Almonds = SR: 0.052*, LR:0.387* Wines= SR:2.096*, LR: 2.736* Grapes=SR: 0.467*, LR: 0.778*

Singapore=0.055* UK=0.016*

Milk = 0.0018 Juice = 0.14* Soft drink =-0.115* Coffee & Tea =-0.0085

Estimated cross promotion elasticities

Apples:

Almonds = SR: 0.221*, LR:-2.31* Wines = SR:0.603*, LR: -0.369* Grapes = SR:0.362*, LR: 5.414*

Almonds:

Apples = SR: 1.028*, LR: 3.775* Wines = SR: -0.062*, LR: 0.027* Grapes = SR: 0.706*, LR: 4.898*

Wines:

Apple = SR: 1.128*, LR: 14.133* Almonds = SR: 0.045*, LR: -1.769* Grapes = SR:0.853*, LR:80.848

Grapes: Apples = SR:0.647*, LR:-4.258* Almonds = SR:0.029*, LR:0.198* Wines = SR:0.067*, LR:0.019*

Singapore:

Australia =-0.011 New Zealand =0.079* Chile =0.601* France =0.186* China = -0.095 Rest of the word =0.048*

UK:

Australia =-0.008 New Zealand =-0.025* Chile =0.045* South Africa =0.034* France =0.079* Rest of the world =0.015

Milk:

Juice =0.0479* Soft drink =-0.0117 Coffee & Tea = -0.0442*

Juice:

Milk = 0.0267* Soft drink = 0.0119 Coffee & Tea =-0.2849*

Soft drink:

Milk =-0.0181 Juice =0.0329 Coffee & Tea =0.2954*

Coffee & Tea: Milk =-0.02* Juice =-0.2304* Soft drink =-0.0864*

13

14

Xiao, Kinnucan, Kaiser (1999) examined the effects of advertising on U.S. non-

alcoholic beverage demand using a two-stage Rotterdam model, which is a major rival

of AIDS model, with annual time-series data covering the period 1970-94. Four

equations of the budget share (milk, juices, soft drinks and coffee and tea) expressed

as an average of the current and preceding year’s budget shares were simultaneously

estimated with advertising expenditure variable of each beverage included as an

explanatory variable. Results suggested many of the cross-advertising elasticities were

larger in absolute value than the own-advertising elasticities, and in some cases exceed

price and income elasticities. The estimated own-advertising elasticity for juices was

positive and significant at 0.134 and for soft drink was negative at -0.115. The

estimated own-advertising elasticity for milk was positive, but not large (0.002) and

not statistically significance. Over all, coffee and tea was the most affected by other

beverages’ advertising even though its own-advertising did not give significant impact,

and milk the least.

None of these studies using demand systems simulated the model system and

calculated a net return of the direct and total effect.

15

CHAPTER 3

MEASURING THE EFFECITIVENESS OF U.S. RICE

EXPORT PROMOTION PROGRAM

3.1 Introduction

Since 1984, the U.S. government has used export promotion as a means to

increase rice exports to other countries. Three main types of government programs

have been used to encourage U.S. rice exports expansion. First, the United States sells

rice on concessional credit terms and donates rice to needy countries bilaterally or

through the World Food Program (P.L. 480, Section 416(b), Food for Progress).

Second, the USDA provides export credit guarantees for commercial financing of U.S.

agricultural exports (GSM- 102, GSM-103, and Supplier Credit Guarantee Program).

Finally, the USDA funds the creation, expansion, and maintenance of foreign markets

for U.S. rice through its market development programs (Market Access Program

(MAP), and Foreign Market Development Program (FMD)). Several other programs

such as the Emerging Market Program, the Qualities Samples Pilot Program, the

Cochran Fellowship Program, and Section 108 also provide assistance to U.S. rice

exports.

The focus of the research reported here is on the effectiveness of the market

development programs for U.S. rice exports. MAP and FMD are partnership programs

between FAS/USDA and numerous non-profit private sector commodity and regional

associations. The oldest market development program is the Foreign Market

Development Program (FMD), also known as the cooperator program. Created in

1955, FMD embodies the agency’s primary goal to develop, maintain and expand

long-term export markets for U.S. agricultural products. Under the partnership,

FAS/USDA and the cooperators pool their technical and financial resources to

16

conduct market development activities outside the United States. These cooperators

compete for USDA funding annually based on their proposed activities. Groups which

have been successful in expanding export markets and contribute a substantial amount

of their own funds are more likely to get increased funding. Under FMD, only generic

promotions are funded. There were 32 cooperator groups which received funding in

1997, covering a broad range of agricultural commodities. One-half of the $27.5

million in U.S. government funding under FMD was spent on feed grains, wheat and

soybeans. Most of the remaining funds went to forest products, meat, rice and poultry

exporters.

The MAP program was originally started in 1985 as the Targeted Export

Assistance Program (TEA), which was created from the 1985 Farm Bill, to offset the

adverse impact of unfair trade practices of other countries on exporters of U.S.

agricultural products. The 1990 Farm Bill replaced TEA with the Market Promotion

Program (MPP). The 1990 Act shifted the focus of the effort from compensating

applicants for unfair trade to a goal of increasing U.S. agricultural exports in

promising foreign markets. The 1996 Farm Bill renamed this program the Market

Access Program (MAP). Like FMD, the MAP program also helps U.S. producers,

exporters, private companies and other trade organizations finance promotional

activities for U.S. agricultural products. While it funds many of the same types of

activities as the FMD program, MAP is intended to be used for shorter- term,

consumer oriented promotions. It is primarily used for high value and processed

products, while FMD is generally used for bulk products. Branded promotions are

permitted under the MAP. Both MAP and FMD use funds from USDA’s Commodity

Credit Corporation (CCC). USA Rice Federation (USARF) and USA Rice Producer

Association (USRPA) are the only two money recipients for rice export promotion.

17

Those two public-private cost-share programs generate over $4 million

annually to promote US rice in all varieties. For 2000, FAS supported USA Rice

Federation through MAP by nearly $2 million and FMD by almost $1.8 million,

accounting for 2.2% and 5.2% of total program allocation, respectively (Wang). The

USA Rice Federation has supplemented its request for nearly $6 million in MAP and

FMD funds with an additional $1.9 million in special programming requests. In 2004,

MAP funding for the U.S. rice industry topped $2.9 million, and reached $4.7 million

in 2006. However, the FMD reduced the allocation from $1.8 million to $1.7 million

in 2004 and decreased it further to $1.46 million in 2006. International rice promotion

focuses on educating foreign consumers about the nutrition of rice, emphasize the

high quality, versatility and dependability of United States rice. The FMD program

has a statutory baseline total allocation for all agricultural commodities of around $34

million. MAP funding, however, is legislated to grow over the life of the Farm Bill,

reaching the 1990 Food, Agriculture, Conservation and Trade Act level of $200

million in 2006. The proportion of money allocated for the U.S. rice industry has

declined over time. U.S. rice exports decreased in world market share from 20% of

global rice exports in 1980s to 14% in 1990s and only 12% in the current decade

(Patricio Mendez del Villar).

There are three objectives of this research. The first objective is to empirically

measure the responsiveness of U.S. export demand with respect to U.S. rice export

promotion. The second objective is to evaluate the overall effectiveness of the

promotion programs in terms benefits relative to costs. Finally, the optimality of

export promotion expenditure levels is investigated by computing a marginal benefit-

cost ratio.

18

3.2 Previous Research

There have been numerous studies that have investigated the impact of export

promotion programs on demand for U.S. agricultural commodities. Table 3.4 provides

a synopsis of 16 export promotion studies in terms of key assumptions, techniques

used, and results. The studies are organized by commodity being studied. While not

exhaustive, the studies depicted in this table provide a good representation of the

literature. The majority of these studies have been commodity and individual country

specific, e.g., U.S. raisin export promotion in Japan. These partial equilibrium studies

have examined the direct (and in some cases indirect) impacts of FAS programs on

specific commodity demand, and have varied in the type of estimation techniques

used (e.g., single equation vs. system of equations estimation), functional form

specification (e.g., linear vs. logarithmic), and variables included in the model.

The majority of studies have employed single-equation estimation. One

problem with this approach is that all explanatory variables are assumed to be

exogenous, however, the own price variable may in fact be endogenous (i.e., be

dependent on the level of per capita imports). Researchers who have assumed price is

exogenous have sometimes justified this by noting that importers are generally price

takers and hence price can be considered as being exogenous. However, this is only

true in the case where the U.S. is a small country in terms of its exports. Some of

these studies have conducted statistical tests of the price endogeneity assumption as

well, and justified this assumption by confirmation of price being exogenous based on

the statistical test. There have been some studies that have addressed the issue by

using statistical estimation techniques that correct for price endogeneity such as an

instrumental variable approach.

While U.S. export promotion has been the focus of these studies, to accurately

measure the impact of promotion on demand, one must sort out the impact of all other

19

factors that impact the demand for the U.S. product. Ideally, an import (or export)

demand model would include own price, price of substitutes and complements,

income, exchange rates, population, trade barrier measures, own export promotion,

and competing export promotion. Unfortunately, it is not always possible to get data

for all these factors, e.g., competing promotion from other countries is usually very

difficult to get and has not been included in any study known to us. In other cases,

studies have been forced to use only partial measures for important variables such as

U.S. export promotion expenditures. Many studies have only included FAS

expenditures for this variable since private industry promotion data were unavailable.

In this case, the estimated promotion impacts on demand could be biased upwards.

While a statistically significant and positive promotion elasticity is a necessary

condition for export promotion to be profitable, it is not a sufficient condition. For

promotion to be profitable it must increase the exported product’s price sufficiently to

cover the per-unit cost of producing, shipping, and promoting the commodity in the

foreign market. Hence, the essential task of the economist in measuring returns to

export promotion is determining the price effect of the demand shift. This requires

specification of the supply side of the market. If this supply curve for any of the

markets in question is horizontal, there can be no price effect from the promotion and

thus no benefit to U.S. producers in terms of increased producer surplus. Thus, an

implicit assumption of export promotion evaluations is that the U.S. accounts for a

sufficiently large share of the total supply in the target market that promotion-induced

shifts in the demand curve for the U.S. product will affect the market price.

Once the price effect of promotion has been determined, the benefit to

producers is calculated using the change in producer surplus. To get a benefit-cost

ratio (BCR), the gain in producer surplus is divided by some measure of promotion

cost, typically total outlays, although some studies separate government from industry

20

costs to distinguish between private and public returns. The majority of studies have

computed benefit-cost ratios by simulating an export demand model without regard to

the supply side of the market. The problem with such studies is that price is implicitly

assumed to be constant, which means there can be no change in producer surplus. In

essence, these studies equate gains in export revenue with gains in producer surplus,

which is appropriate only when supply is fixed. Recalling that the area under the

supply curve represents total costs, it is clear that when a demand curve shifts along a

horizontal supply curve, the revenue gain is exactly matched by an increase in

production cost and thus the profit or surplus gain is zero. Conversely, when demand

shifts along a vertical supply curve there is no increase in production cost (since

quantity is fixed) and thus the revenue gain represents pure profit or economic rent. In

an export context, the outward shift in the export demand curve along a horizontal

import supply curve merely diverts quantity from the U.S. market with no change in

U.S. producer revenue. Stated differently, the revenue gain in the export market is

exactly matched by a revenue loss in the domestic market. This issue, which is

discussed in some detail in Kinnucan and Mryland, means that benefit-cost estimates

from the literature must be interpreted with caution.

The majority of studies reviewed here shows a positive relationship between

export promotion expenditures and export demand. For example, of the 15 studies in

Table 3.4 done on individual commodities, eight studies found statistically significant

positive export promotion effects for all countries and commodities in their specific

study. The remaining seven studies obtained mixed results, with some commodities

and/or importing countries studied showing no measurable demand effect. None of

the studies presented evidence to suggest non-price export promotion has no effect on

export demand.

21

What is less clear from the literature review is the economic impact of export

promotion. Although many studies report benefit-cost ratios, most of these are based

on revenue gains rather than gains in producer surplus, the preferred metric. Indeed,

of the 16 studies in Table 3.4, only two incorporated a supply response in the

simulation of benefit-cost ratios. Also, the majority of studies estimated marginal

rather than average BCRs. In determining the private or social returns to promotion

the relevant metric is the average BCR as this measures the return to all dollars

expended, not simply the last dollar. Marginal BCRs are more appropriate for

determining whether the level of promotion is too low or high relative to the economic

optimum. Of the 16 studies listed in Table 3.4, only four estimated an average BCR.

Finally, as emphasized by Jakus, Jensen and Davis, few studies provide confidence

intervals, and thus one does not know whether reported benefit-cost ratios are truly

greater than one.

In the current study, a single equation, instrumental variable regression

approach is used to estimate export demand for U.S. rice. The constant price

(horizontal supply function) assumption is relaxed by introducing a constant elasticity

excess supply equation and the average BCRs is computed for a range of assumed

own price elasticities of excess supply. Confidence intervals for the BCR are also

computed and reported. Finally, marginal BCRs are estimated in order to explore the

optimality of rice export promotion. The detailed methods used to evaluate rice export

promotion in this study are next addressed.

3.3 Econometric Model

In this study, an export demand equation for U.S. rice (USEXt) is estimated in

logarithmic form using annual data from 1984 through 2005. Ideally, a longer time

series would be preferred, however, there was no U.S. rice export promotion program

22

prior to 1984. The following export demand determinants are included to ascertain

their impacts, if any, on annual export demand for U.S. rice 2: deflated price of U.S.

milled rice exports (USMILPWt), deflated prices of Thai and Vietnamese milled rice

exports (THMILPWt, VIETMILPWt respectively), the summation of deflated GDP of

major U.S rice importers 3 (GDPSUMWt) and U.S. rice export promotion

expenditures 4(EXPROWSt) deflated by world price index and multiplied by Special

Drawing Rights (SDR) 5.

The Thai and Vietnamese rice export prices are included because these two

countries are the major rice exporters in the world and the chief competitors to U.S

rice in different markets. The relationship between the competitors’ deflated prices

should be positive because they are a substitute for U.S. rice. The summation of the

GDP of major U.S rice importers should have a positive impact on U.S. rice exports.

Unlike previous studies, which only consider the USDA/FAS expenditures for U.S.

rice export promotion due to lack of data, e.g. Dwyer (1995), Wang (2005), this

research uses combined USDA/FAS- MAP and FMD expenditures as well as private

cooperator expenditures to measure the total promotion impact. Because previous

research (e.g. Kinnucan, 1982, 1983, 1985; Le, Kaiser, and Tomek, 1998; Pritchett,

Liu, and Kaiser, 1998; Schmit and Kaiser, 1998) has indicated that the carryover

effect of promotion is generally less than one year, the annual rice model does not

2 The following data sources are used for the variables: volume of U.S. rice exports comes from FAS, USDA and IRRI; export unit value for rice-paddy milled in the U.S., Thai and Vietnamese rice comes from FAO; the summation of GDP of major U.S rice importers comes from The Economist Intelligence Unit; annual export promotion expenditures from FAS, USDA; and SDR from IMF. 3 The summation of GDP for Middle East and Africa countries include: Iran, Iraq, Saudi Arabia, Turkey, Jordan, South Africa, Ghana, and Nigeria, Latin Americas and Caribbean - including Mexico, Brazil, Peru, Haiti, Cuba, Costa Rica, Nicaragua, Jamaica, Republic of Dominican, and Honduras, Canada, Japan, South Korea, EU-25. 4 The export promotion expenditure levels partially depend on exchange rates since, for example, when the U.S. dollar devalues, export promotion expenditures will have lower purchasing power. 5 Created by the IMF in 1969, the SDR was redefined as a basket of currencies, today consisting of the Euro, Japanese Yen, Pound Sterling, and U.S. dollar. It is calculated as the sum of specific amounts of the four currencies valued in U.S. dollars, on the basis of exchange rates quoted at noon each day in the London market.

23

include lagged export promotion. Furthermore, preliminary regression with one and

two year lags resulted in statistically insignificant coefficients for the lagged variables.

To address the problem of price endogeneity, an instrumental-variable

regression approach is used, where the U.S. export price is regressed on a set of

exogenous variables. Hence, the model consists of two equations: equation 3.1: a price

equation 6 to be used as an instrumental variable for the endogenous U.S. export price,

and equation 3.2: the demand equation for U.S. rice exports:

(3.1) USMILPt = α0 + α1GDPSUMt + α2 WPIt + α3 SDRt + α4 EXPROt

+ α5 USMILPt-1 + α6 THMILPt + α7 VIETMILPt

(3.2) log(USEXt) = β0 + β1 log(USMILPWIVt) + β2 log(THMILPWt)

+ β3 log(VIETMILPWt) + β4 log(GDPSUMWt )

+ β5 log(EXPROWSt ),

where WPI is the world price deflator, SDR is special drawing rights, and

USMILPWIV is the predicted value from equation 3.1, deflated by world price index.

3.4 Econometric Results

The estimated price instrumental variable equation is:

(3.3) USMILPt = – 250.353 + 0.000028 GDPSUMt – 1.435 WPIt + 144.266 SDRt (-2.28) (1.39) (-2.15) (1.31)

+ 0.00000445 EXPROt + 0.740605 USMILPt-1 (3.401) (7.43)

+ 0.188102 THMILPt + 0.687075 VIETMILPt (0.95) (4.82) R2 = 0.79; Durbin-h = -1.24,

6 All exogenous variables used for a price equation are nominal (not deflated by world price index

(WPI) nor multiplied by special drawing rights (SDR))

24

where the values in parentheses under the coefficients are their respective t-values, R2

is the adjusted coefficient of determination. The instrumental equation fits the data

well with an adjusted coefficient of variation of 0.79. The Breusch-Godfrey’s serial

autocorrelation test identified statistically significant high-order autocorrelation7. An

AR (3) process was used to remove autocorrelation from the error structure. Durbin-h

statistic was -1.24, which is less than a Durbin-h critical value at the 5% significant

level (-1.96). Thus we fail to reject the null hypothesis that there is no-autocorrelation.

The predicted value for this equation is used as an instrument for the U.S. rice price

for equation 3.4.

The estimated export demand equation for U.S. rice (equation 3.2) is:

(3.4) log (USEXt) = 3.425 – 1.028 log(USMILPWIVt) + 0.637 log(THMILPWt) (1.71) (-3.57) (2.86)

+ 0.461 log(VIETMILPWt) + 1.047 log(GDPSUMWt ) (1.84) (5.90)

+ 0.143 log(EXPROWSt ) R2 = 0.80 , DW=2.29

(2.85)

Because autocorrelation is detected, a moving average error term is appended

to the regression, and the resulting equation is subsequently free from autocorrelation.

The equation fits the data well; the adjusted R-squared indicates that the equation

explains about 80% of the variations in U.S. rice export demand. It has elasticity

signs, which are consistent with economic theory, and all estimated coefficients are

statistically significant at conventional confidence levels (95% and 90%). No

multicollinearity was detected in the equation.8

7 The Bresuch-Godfrey’s serial autocorrelation test identified a high-order autocorrelation which is at 10% significant level with p-value 0.06. 8 Variance Inflation Factor (VIF) is employed in this case to test multicollinearity. It measures the impact of collinearity among the X's in a regression model on the precision of estimation. It expresses the degree to which collinearity among the predictors degrades the precision of an estimate. Typically a VIF value greater than 10 is of concern. In this paper, VIF, on average, is 2.91 which indicate no multicollinearity.

25

The estimated equation indicates that the price of U.S. rice is an important

factor in explaining annual variations in its own rice export demand. The estimated

own-price elasticity is – 1.028. Unlike domestic demand, it is common to find elastic

own price elasticities for export demand (e.g., Kaiser, Liu, and Consignado found

elastic price elasticities for export raisin demand).

The prices of rice exports from other countries, which compete with the United

States, are also found to be important factors in the equation. The cross-price elasticity

of U.S. rice export demand with respect to Thai and Vietnamese rice export prices are

estimated at 0.637 and 0.461, respectively. These results suggest that while both

countries are major competitors of the United States, Thailand is slightly more

important in terms of impacting export demand for U.S. rice.

The summation of deflated GDP of the major U.S. rice importers is an

important determinant of the export demand for U.S. rice. The coefficient associated

GDP in countries, which are top U.S. rice importers, is 1.047, indicating the U.S., rice

is a normal good.

Finally, the coefficient associated with the rice export promotion variable is

positive and statistically different from zero. The statistical evidence supports the

notion that U.S rice export promotion programs, which are a public-private

contribution, have the effect of increasing the export demand for its rice. The

estimated export promotion elasticity is 0.143.

3.5 Simulation Analysis

Based on the econometric results, it is clear that rice promotion expenditures

have a positive and statistically significant impact on U.S. rice exports to the world

market. Next, the estimated equation is simulated to address the remaining objectives

of this study. Two scenarios are of interest: (1) a baseline scenario with U.S. rice

26

export promotion expenditures at historical levels – export promotion programs are in

effect; and (2) a counterfactual scenario with no U.S. rice export promotion. In the

first scenario, all rice export demand determinants are set equal to their historic levels.

The second scenario is identical to the first, except U.S. rice export promotion

expenditures are set to a small amount.9 The difference between the two scenarios

gives the total impact of the U.S. rice export promotion programs on the U.S. rice

exports to all trading partners.

Figure 3.1 illustrates the simulation result for U.S. rice exports. From 1984-

2005, FMD and MAP expenditures increased U.S. rice exports by 24,269,361.55

metric tons in total, or an average of 1,103,152.798 metric tons per year. Hence, U.S.

rice export promotion has had a very large impact on total U.S. rice exports.

Figure 3.1: Simulated U.S. Rice Exports with and without Export Promotion.

9 Due to the logarithmic functional form, export promotion expenditures are set proportionally to 2 percent of historic level of export promotion expenditure in this scenario, assuming no export promotion programs because log of zero is undefined.

0

500000

1000000 1500000 2000000 2500000

3000000 3500000 4000000

4500000

5000000

1984 1989 1994 1999 2004

Metric tons

Base Scenario No program

27

3.5.1 Average Benefit-Cost Ratio

While these results indicate a positive impact of export promotion on U.S. rice

exports, what remains a key concern is the impact promotion has on industry producer

surplus compared with promotion costs. The increase in export demand due to export

promotion described above assumes that all other demand determinants, including

price, would remain constant. However, as argued earlier, generally an increase in

demand will cause price to increase as well, provided that the demand increase is not

perfectly offset by an increase in quantity supplied (as in the infinite supply response

case). Hence, in order to evaluate the full effect of U.S. rice export promotion

programs on quantity and price, one needs to incorporate an excess supply response

for U.S. rice into the model.

An approach similar to that by Alston et al. (1996) is followed. In this

approach, the excess supply response is incorporated using a constant elasticity form,

and sensitivity analysis is conducted on a range of assumed own-price supply

elasticities. The simulation procedure begins on the export demand side, where

predicted quantities of rice export demand (QtD) are simulated from the estimated

export demand equation. Then, using the procedure discussed by Alston et al. (1996),

excess supply is defined in constant elasticity form and equated with the predicted

export demand quantities. Changes in export demand due to U.S. rice export

promotion then affect the level of production and price. Specifically, the excess

supply function is defined as:

(3.5) QtES = At Pt

є,

where At = QtD / Pt

є and є is the price elasticity of excess supply. Pt is the U.S. export

rice price in year t. At is varied from year to year to ensure that, given the actual

28

values of prices and the other export demand determinants each year, the excess

supply equation passes through the point defined by the predicted quantity from the

export demand model and actual U.S. rice price. The change in net benefits due to

export promotion programs is computed for each year from 1984 to 2005 as the

difference in producer surplus (VPS) between the two scenarios, which

mathematically is equal to

(3.6) VPSt = ( PtQt - Pt’Qt’) / (1+ є),

where PtQt represents the scenario with export promotion programs and Pt’Qt’

represents the scenario without export promotion programs.

By equating the equations for excess supply and export demand and solving

for world market equilibrium, U.S. actual prices and predicted quantities (from the

export demand model) are obtained given the historical values for the explanatory

variables. The counterfactual scenario is then simulated assuming no export

promotion expenditures by setting its value close to zero. Excess supply elasticities

are chosen between 1 and 10 to examine a broad range of benefit-cost ratios.

Table 3.1 presents the average annual impacts and BCRs (from 1984 to 2005)

for U.S. rice export promotion efforts for the various assumed own-price elasticities of

excess supply. U.S. rice export promotion has had a positive impact on the U.S rice

export price over this period under all supply response scenarios. The average

increase in price ranges from $84.21, in the case of the most inelastic supply response

(ε= 1.0), to $17.28 per ton, in the case of the most elastic supply response (ε= 10.0).

The reason the positive price impacts become lower as the assumed supply response

gets larger is that under the larger supply response scenarios, producers are dampening

the positive price impacts of the increased export demand by increasing quantity

29

supplied to the world market relative to the lower supply response scenarios. The

average impact over all supply responses is $36.33 per ton. In other words, had there

not been U.S. rice export promotion, the average rice export price would have been

$36.33 per ton lower from 1984 to 2005 than it actually was.

U.S. rice export promotion has had a positive impact on producer surplus over

this period as well. The average increase in producer surplus due to export promotion

range from $204.37 million per year, in the case of the least elastic supply response

(ε= 1.0), to $37.5 million per year, in the case of the most elastic supply response (ε=

10). The reason for the negative relationship between supply elasticities and producer

surplus is identical to that described above for supply elasticities and price. The

average increase in producer surplus over all supply responses is $83.25 million per

year. Hence, it is clear that U.S. export promotion has had a significant and positive

impact on industry profits since 1984.

How does the gain in producer surplus compare with the costs of export

promotion? To answer the question, an average benefit-cost ratio is computed (see the

bottom row of Table 3.1). The average BCR exceeds 1.0 for every supply response

considered in the simulation. For the least elastic supply response (ε= 1.0), the

average BCR is 30.05. This implies that, on average over the period 1984-2005, the

benefits of the promotion programs have been over 30.05 times greater than the costs.

At the opposite end of the spectrum in supply response (ε = 10), the average BCR is

computed to be 5.51, implying that the benefits of the USHBC are 5.51 times greater

than the costs. Given the wide range of supply responses considered in this analysis,

and the fact that the BCR is above 1.0 in all cases, there is significance evidence that

U.S. rice export promotion has been profitable for the U.S. rice industry. The average

BCR over all supply responses is 12.20, i.e., the benefits of U.S. rice export promotion

exceed the costs by 12.2.

30

Questions often arise about the accuracy of these BCR estimates in economic

evaluations of commodity checkoff programs. BCRs are generally large because

promotion expenditures are very small relative product value, and therefore only a

small demand effect is needed to generate positive and large returns. For example,

average U.S. rice export promotion expenditures in 2005 were less than 0.013 percent

of the farmer value of rice marketing. Still, this relatively small investment in U.S.

rice export promotion increased producer surplus by over $83.25 million per year

since 1984 (average of all excess supply elasticities). The resulting benefit-cost ratio is

therefore quite large.

Table 3.1: Average annual world market impacts and benefit-cost ratios due to U.S. rice export promotion, 1984-2005

To make allowances for the error inherent in any statistical estimation, a 95

percent confidence interval is calculated for the above average BCRs. Table 3.2

presents the lower bound on the BCR for the 95 percent confidence interval. The

estimated lower bound of the average BCR for the lowest assumed supply response

for the period 1984-2005 is 15.09. This result demonstrates that one could be

confident 95 percent of the time that the true average BCR for this assumed supply

response is not lower than 15.09.The lower 95 percent confidence bound for the

average BCR in the highest assumed supply response for this period is 2.79. Hence, it

Own-Price Elasticity of Excess Supply

1 2 3 4 5 6 7 8 9 10

Change in Producer Price ($/ton) 84.2 58.9 45.3 36.7 30.9 26.7 23.5 20.9 18.9 17.3

Change in Producer Surplus(million $) 204.4 136.7 102.7 82.3 68.6 58.8 54.5 45.8 41.2 37.5

Change in Promotion Cost (million $) 7.87 7.87 7.87 7.87 7.87 7.87 7.87 7.87 7.87 7.87

Benefit-Cost Ratio 30.05 20.11 15.11 12.1 10.1 8.65 7.57 6.73 6.06 5.51

31

is reasonable to conclude that the above confidence lower bound gives credence to the

previous finding that the benefits of the promotion programs have been considerably

greater than their cost.

Table 3.2: Lower-bound 95 percent confidence interval for benefit-cost ratios due to U.S. rice export promotion, 1984-2005

3.5.2 Marginal Benefit-Cost Ratio

In order to explore the “optimality” of the rice export promotion expenditure levels,

marginal simulation analysis is also conducted. The estimated demand equation is

used to simulate the outcome of an additional scenario and the results are compared

with the baseline scenario. In the third scenario, export promotion expenditures are

increased by 1 percent above baseline scenario levels, while all other exogenous

variables are the same as before.

Table 3.3: Marginal benefit-cost ratios a due to rice export promotion, 1984-2005

Own-Price Elasticity of Excess Supply

1 2 3 4 5 6 7 8 9 10

Benefit-Cost Ratio – Lower Bound

15.09

10.04

7.52

6.01

5.00

4.29

3.75

3.34

3.00

2.79

Own-Price Elasticity of Excess Supply

1 2 3 4 5 6 7 8 9 10

Changeb in Producer Price ($/ton) 0.67 0.45 0.34 0.27 0.12 0.10 0.09 0.08 0.07 0.06

Changeb in Producer Surplus(million $) 0.67 0.45 0.34 0.27 0.23 0.19 0.17 0.15 0.34 0.12

Changeb in Promotion Cost (million $) 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08

Marginalb Benefit-Cost Ratio 9.61 6.44 4.84 3.88 3.23 2.77 2.43 2.16 1.94 1.76

a The marginal impact of a 1 percent increase in export promotion expenditures on U.S. rice exports.

b The change in producer prices, producer surplus, promotion costs and marginal benefit-cost ratio in a particular own-price elasticity of excess supply is an average from 1984 to 2005.

32

The estimated marginal BCRs are presented in Table 3.3. The marginal BCRs

are well above 1.0 for every assumed own price elasticity of supply. For example, for

the least elastic supply response (ε = 1.0), the marginal BCR is 9.61 indicating that an

incremental $1.00 increase in promotion would yield $9.61 in producer surplus to the

rice industry. For the most elastic supply response (ε= 10.0), the marginal BCR is

1.76, which is still well above 1.0. These high marginal BCRs imply that the rice

industry should explore the option of raising more money for export promotion since

doing so would return benefits that are substantially greater than costs. These high

marginal BCRs are also very common among agricultural checkoff programs, which

indicate that these collective action programs are under-funded from an economic

optimality point of view (see Table 3.4 for marginal BCRs for other commodities).

3.6 Conclusions

The objectives of the study were to measure the responsiveness of U.S. rice

export demand with respect to U.S. rice export promotion, evaluate the overall

effectiveness of the promotion programs, and investigate the optimality of export

promotion expenditure levels.

An econometric export demand equation in double logarithmic function was

estimated to measure the export promotion elasticity, while controlling for other

demand determinants such as own price, price of competing countries, income, and

exchange rates. The results support the hypothesis that the export promotion programs

have had a positive impact on the U.S. rice export demand. The export promotion

elasticity was computed to be 0.143, which was statistically significantly different

from zero.

33

Table 3.4: Key results from economic impact studies on U.S. export promotion.

California raisins

US orange juice

US orange juice US fresh grapefruit US apples US apples CA table grapes

US frozen potatoes

Study

Kaiser, Liu, and Consignado (2005)

Lee and Brown (1986)

Armah and Epperson (1997)

Fuller, Bello, and Capps (1992)

Rosson, Hammig, and Jones (1986)

Richards and Patterson (1997)

Alston et al. (1997)

Lanclos, Devodoss, and Guenthner (1997)

Activities evaluated

Industry and FAS programs

Three Party program

Industry and FAS programs

FAS Three Party and TEA programs

Industry and FAS programs

Industry and FAS programs

Industry and FAS programs

Industry and FAS programs

US export promotion in:

Japan and UK 13 European countries

France, UK, Germany, Japan Netherlands

Japan, Canada, France, and Netherlands

All countries US has programs in

Singapore and UK

Asian countries

Japan, Mexico, Philippines, Thailand

Period of estimation

1965-98 1973-82 (panel data)

1984-92 (panel data) 1969-88 quarterly 1972-81 1962-93 1976-94 1978-93

Type of model

Import demand, single equations

Import demand, single equations

Export demand, single equation

Import demand, single equations

Export demand single equations

Import demand, LES/AIDS demand systems

Single equation, export demand

Import demand, single equations

Estimated promotion elasticities

Japan=0.029* UK=0.133*

Promotion elastitcities not given

France=0.014 Germany=0.044* Japan=0.014 Netherlands=0.302* UK=0.014*

Japan=0.109* Netherlands=0.153* France=0.234*

Apples=0.51* Singapore=0.055* UK=0.016*

0.21*

Third Party: Japan=0.03* Philippines=0.53* Thailand=0.87*

Estimated benefit-cost ratio

Japan: AGBCR=5.13 MGBCR=0.42 UK: AGBCR=15.29 MGBCR=3.19

For all countries, MGBCR=5.51

MGBCRs: France=7.44 Germany=37.10 Japan=5.61 Netherlands=51.92 UK=7.64

MGBCR:Japan=5.02 Netherlands=6.65 France=4.13 Canada=no promotions

MGBCR=60.0 NA

ABCR: 4.1-9.4 MBCR: 4.1-4.2

Third Party MGBCRs: Japan=1.29 Philippines=11.77 Thailand=16.36

Assumed no promotion price impact

Yes Yes Yes Yes Yes NA No Yes

Peer reviewed Yes Yes Yes Yes Yes Yes Yes Yes

33

34

Table 3.4 (continued)

US pecans US walnuts US almonds US cotton US red meat US red meat US soybeans All US food exports

Study

Onunkwo and Epperson (2000)

Weiss, Green, and Havenner (1996)

Halliburton and Henneberry (1995)

Solomon and Kinnucan (1993)

Le, Kaiser, and Tomek (1998)

Comeau, Mittelhammer, and Wahl (1997)

Williams et al. (1998)

Dwyer (1995)

Activities evaluated

Industry and FAS programs

Industry and FAS programs

FAS FMD and MPP programs

FAS programs FAS FMD and TEA programs

FAS MPP and TEA programs

Industry and FAS programs

FAS programs

US export promotion in:

Asia and EU Japan

Japan, Taiwan, Hong Kong, Singapore, South Korea

6 countries in the Pacific Rim

S. Korea, Taiwan, Hong Kong, Singapore

Japan EU, Japan, and Rest of the World

World

Period of estimation

1986-96 (panel data)

1986-96 (monthly data)

1986-92 (panel data)

1965-85 1984-94 (panel data)

1973-94 1969-96 1975-92

Type of model

Export demand, single equation

Event analysis Import demand, single equations

Armington trade model

Import demand, single equations

Inverse Almost Ideal Demand System

SOYMOD world market model

Armington trade model

Estimated promotion elasticities

Asia=0.98* EU=0.06*

$1000 in promotion increased exports by 4.5 tons

3 models range from -0.2788 to 0.85

Japan=0.53* S. Korea=0.045* Hong Kong=0.21* Philippines=0.26* Thailand=0.045 Taiwan=-0.54

Korea=0.598* HK=-0.019 Taiwan=0.047 Singapore=0.034

Japan price flexibilities wrt promotion ranged from 0.11* to 0.128*

Soybeans: EU=0.0234* Japan=0.0367* ROW=0.068* Soymeal: EU=0.0445* Japan=0.0733* ROW=0.0516* Soyoil: EU=0.0446* Japan=0.0323* ROW=0.0156*

Short-run=0.0135* Long-run=0.15*

Estimated benefit-cost ratio

Asia: MGBCR= 6.45 EU: MGBCR= 6.75

MGBCR=5.85

MGBCRs: Japan=4.95 Taiwan=5.94 HongKong=3.69

NA MGBCR= 15.62 to 47.32 for all 4 countries

MGBCR for beef ranged from 15.56 to 18.11

ABCR: 13.5 (1978-89) 5.3 (1990-94) 11.3 (1978-94)

AGBCR=16.0

Assumed no promotion price impact

Yes Yes Yes NA Yes Yes No Yes

Peer reviewed Yes No Yes Yes Yes Yes Yes No

34

35

Furthermore, simulation procedures were employed to estimate the magnitude

of the impact of export promotion programs on total rice exports. The simulation

indicated that U.S. rice export promotion programs increased U.S. rice export

quantities by an average of 1,103,152.798 metric tons per year. Thus, we can conclude

that U.S. rice export promotion programs have been effective in enhancing foreign

market demand for U.S. rice.

Average benefit-cost ratios were also computed for U.S. rice export promotion

based on a range of excess supply own price elasticities (from 1 to 10). The BCRs

ranged from 30.05 for the most inelastic estimate, to 5.51 for the most elastic. Since all

of these values were well above 1.0, the main policy conclusion was that the benefits

of export promotion programs in terms of enhancing producer welfare have been

much greater than the costs of the programs.

Finally, marginal simulation analysis was conducted to explore the optimality

of rice export promotion. The marginal BCRs over the period of study were estimated

between 9.61 and 1.76 given the own-price elasticity of excess supply between 1 and

10. Thus, all results indicate that the U.S. is under-investing in rice export promotion

from an optimality standpoint.

36

REFERENCES

Alston, J.M. and Carman, H.F. and Chalfant, J.A. and Crespi, J.M. and Sexton, R.J.

Evaluation of Prune Promotion by the California Dried Plum Board. In: Kaiser,

H.M., Alston, J.M., Crespi, J.M. and Sexton, R.J. (eds). The Economics of Commodity

Promotion Programs: Lessons from California. Peter Lang, New York (2005).

Chapter 6. pp. 109-136.

Alston, J.M. and Chalfant, J.A. and Christian, J.E. and Meng, E. and Piggott, N.E. (1997). The

California Table Grape commission’s Promotion Program: An Evaluation.

Berkeley CA: University of California Division of Agriculture and Natural Resources,

Giannini Foundation Monograph 43, November.

Alston, J.M. and Crespi, J.M. and Kaiser, H.M. and Sexton, R.J. Introduction to the

Economics of Commodity Promotion Programs. In: Kaiser, H.M., Alston, J.M.,

Crespi, J.M. and Sexton, R.J.(eds). The Economics of Commodity Promotion

Programs: Lessons from California. Peter Lang, New York (2005). Chapter 1. pp. 3-

12.

Armah, B. and Epperson, J. (1997). Export Demand for U.S. Orange Juice: Impacts of

U.S. Export Promotion Programs. Agribusiness(13). pp. 1-10.

Childs, N.W. and Burdett, A. (2000). The U.S. Rice Export Market. Rice Situation and

Outlook/RCS - 2000/November 2000. pp. 48-54.

Crespi, J.M. and Sexton. R.J. Evaluating the Effectiveness of California Almond

Promotion: How Much Did the Litigation Cost Producers?. In: Kaiser, H.M.,

Alston, J.M., Crespi, J.M. and Sexton, R.J.(eds). The Economics of Commodity

Promotion Programs: Lessons from California. Peter Lang, New York (2005).

Chapter 8. pp. 167-190.

37

Halliburton, K. and Henneberry, S.R. (1995). The Effectiveness of U.S. Nonprice

Promotion of Almonds in the Pacific Rim. Journal of Agricultural and Resource

Economics (20). pp. 108- 121.

Jakus, P.M. and Jensen, K.L. and Davis, G.C. (2003). Revenue Impacts of MPP Branded

Funds: A Firm-Level Analysis. Agricultural and Resource Economics Review (32).

pp. 184-197

Kaiser, H.M. and Liu, D.J. and Consignado, T. (2003). An Economic Analysis of California

Raisin Export Promotion. Agribusiness19(2). pp. 189-201.

Kaiser, H.M. An Economic Analysis of Domestic Impacts of the Walnut Marketing

Board's Marketing Programs. In: Kaiser, H.M., Alston, J.M., Crespi, J.M. and

Sexton, R.J.(eds). The Economics of Commodity Promotion Programs: Lessons from

California. Peter Lang, New York (2005). pp 191-208.

Kinnucan, H.W. and Zheng, Y. (2005). National Benefit-Cost Estimates for the Dairy,

Beef, Pork, and Cotton Promotion Programs. In: Kaiser, H.M., Alston, J.M.,

Crespi, J.M. and Sexton, R.J.(eds). The Economics of Commodity Promotion

Programs: Lessons from California. Peter Lang, New York (2005). Chapter 12. pp.

255-284.

Kinnucan, H.W. and Myrland, φystein (2001). A Note on Measuring Returns to Nonpirce

Export Pormotion. Agribusiness: An Internaional Journals(17). pp. 423-433.

Lanclos, D. and Devadoss S. and Guenthner, J. (1997). Impacts of Foreign Direct

Investment and Advertising on the Export Demand for U.S. Frozen Potatoes.

Agribusiness(13). pp. 273-284

Le, C.T. and Kaiser, H.M. and Tomek, W.G. (1998). Export Promotion and Import

Demand for U.S. Red Meat In Selected Pacific Rim Countries. Agribusiness14(2).

pp. 95-105.

38

Onunkwo, I.M. and Epperson, J.E. (2000). Export Demand for U.S. Pecans: Impacts of

U.S. Export Promotion Programs. Agribusiness: An International Journal(16). pp.

253-266.

Richards, T.J. and Patterson, P.M. (1998). Dynamic Complementarity in Export Promotion

: The Market Access Program in Fruits and Vegetable . Journal of Agricultureal

and Resource Economics(23).pp. 319-337.

Rosson, C. and Hammig, M. and Jones, J. (1986). Foreign Market Promotion Programs:

An Analysis of Promotion Response for Apples, Poultry, and Tobacco.

Agribusiness(2), pp. 33-42.

Schmit, T. and Kaiser, H.M. (1998). Egg Advertising, Dietary Cholesterol Concerns, and

U.S. Consummer Demand. Agricultural and Resource Economics Review(27). pp.

43-52.

Solomon, H. and Kinnucan, H.W. (1993). Effects of Non-price Export Promotion: Some

Evidence for Cotton. Australian Journal of Agricultural Economics (37). pp.1-15.

U.S. Department of Agriculture, Economics and Statistics Service. Rice Situation and

Outlook. Washington DC, 1960-2005.

Wang, Yiqian (2005). An Examination of U.S. Rice Export Promotion Programs. Master

Thesis of The Department of Agricultural Economics and Agribusiness. Louisiana

State University and Agricultural and Mechanical College.

Weiss, K.R. and Green, R.D. and Havenner, A.M. (May, 1996). Walnuts in Japan: A Case

Study of Generic Promotion Under the USDA’s Market Promotion Program. In

Agricultural Commodity Promotion Policies and Programs in the Global Agri-Food

System. Proceedings of NEC-63 Conference. Ithaca, New York. NICPRE, Cornell

University. pp. 47-80.

Williams, G.W. and Shumway, C.R. and Love, H.A. and Ward, J.B. (May,1998).

Effectiveness of the Soybean Checkoff Program. Texas Agricultural Market

39

Research Center Report. Department of Agricultural Economics. Texas A & M

University.

40

CHAPTER 4

THE HALO EFFECTS OF U.S. GRAIN EXPORT PROMOTION

4.1 Introduction

The U.S. government funds export promotion programs for agricultural and

food commodities through partnerships with nonprofit commodity trade associations.

The two most prominent government programs operated by USDA’s Foreign

Agricultural Service are the Market Access Program (MAP) in 1985 and the Foreign

Market Development Program (FMD) in 1955. In the 2002 Farm Bill, these two

programs experienced a large increase in their annual budgets of $250 million.

Currently, annual funding for MAP and FMD exceed $500 million. Because of the

large amount of money invested, these programs have been studied extensively in

terms of their economic impacts (see Alston et al. for a thorough review). Almost all

previous studies have focused on the direct effects of export promotion for a specific

commodity on that commodity’s exports, with the majority finding that promotion has

a statistically significant impact on exports.

However, these programs remain highly controversial with opponents claiming

they amount to "corporate welfare". Proponents contend that export promotion

programs enable U.S. exporters to be competitive in the global market and level the

playing field with other countries who also have government funded export

promotion. From an economic point of view, government’s role in funding export

promotion can be rationalized only if market failure exists. One type of market failure

is positive externalities produced by export promotion of one commodity in terms of

increases in export demand for other U.S. commodities. For example, a trade shows

with the objective of promoting rice may also promote other food grain or feed grain

exports or, perhaps, other entirely different commodities. Since these indirect or “halo

41

effects” are not considered by the private firm doing the promoting, the level of export

promotion is likely lower than optimal from a social point of view (Dwyer).

Unfortunately, there have been few studies that have empirically examined

whether a halo effect actually exists. To our knowledge, there are only two studies

that have explored this issue, and both focused only on four horticultural commodities:

apples, almonds, wine, and grapes. Richards and Patterson (1998) used a dynamic

dual model of U.S. export supply for these four commodities, which included FAS

promotion expenditures (TEA and MAP funds) as explanatory variables. Estimation

was based on time series data from 1984-95 pooled with country-level data for

Mexico, Sweden, Norway, Finland, the United Kingdom, France, Germany, Malaysia,

Singapore, the Philippines, and Japan. The authors estimated both direct and cross

effects of U.S. export promotion for all four commodities, and all estimated elasticities

were statistically significant. The authors found that cross effects were important, and

in some cases were as large as direct effects.10

In terms of the cross effects, the

elasticities were also all statistically significant, and the majority of them were positive

indicating a halo effect among the four commodities.

Global Insight updated the Richards and Patterson study using data from 1985-

2004 with a demand system for the same four commodities. This study found eight of

the cross-promotion elasticities to be positive and statistically significant (compared

with four negative cross promotion elasticities). This is consistent with Richards and

Patterson’s findings regarding the cross promotion effects of U.S. promotion. Similar

to Richards and Patterson, the magnitude of the cross effects were generally larger

10 Richard, Ispelen, and Kagan (R-I-K) (1997) indicated that promotion is more effective the less elastic is demand. The empirical results of the study also show that the effects of export promotion by one country have larger spillover effects the more inelastic is demand for a rival’s product. In other words, the free-rider effect of export promotion will be greatest when other countries are able to successfully differentiate their product. If the product is inherently difficult to brand, the benefit from promotion will flow to those who are most able to achieve some measure of differentiation.

42

than the own effects. However, unlike Richards and Patterson, the cross promotion

elasticities were smaller and more plausible in magnitude, ranging in value from 0.099

to 0.883.

In somewhat related research, Richards, Ispelen and Kagan (1997) examined

whether foreign market promotion constitutes a type of “international public good”.

Export promotion is considered an international public good when it has positive cross

effects that dominate country-specific effects, i.e., halo effect on other countries. In

other words, when promotion impacts total exports to a country, but not a specific

country’s share, then promotion has public good aspects and thus individual countries

have no incentive to promote. Very little research has been conducted on these cross

effects of promotion in an integrated framework. Failing to take these cross effects

into account tends to generate results that overstate returns from promotion and only

partially measure the full promotion impact (Kinnucan and Zheng 2005).

The majority of economic evaluation studies on U.S. export promotion have

relied on the single equation approach, which specifies a single demand equation for a

specific commodity as a function of export promotion and other demand factors. Cross

effects are more properly evaluated using a demand system approach rather than with

single-equation demand models. For instance, a single demand equation for U.S. rice

exports cannot measure the impacts of U.S. rice export promotion on other related

grains and competing rice from other countries. The few studies that have examined

these cross effects have used multi-equation systems as an integrated framework such

as the Rotterdam model, i.e., Xiao, Kinnucan and Kaiser (1999) and the Almost Ideal

Demand Systems (LA/AIDS), i.e., Richard, Ispelen, and Kagan (1997).

The objective of the research reported in this article is to examine both the own

and cross effects of U.S. export promotion for U.S. and non-U.S. grains. We focus on

two previously neglected dimensions of the cross-effects. First, we examine whether a

43

halo effect for U.S. commodities exists. Second, we look at whether U.S. grain export

promotion has public goods characteristics, i.e., does U.S. grain promotion have a halo

effect on grain demand for other countries? We use a linear approximation of an

Almost Ideal Demand System (LA/AIDS) model for U.S. and non-U.S. rice, wheat,

and sorghum that include U.S. export promotion as one of the explanatory variables.

Our goal is to provide further empirical evidence for the halo effect issue by using

U.S. grains as a case study since this sector has not been previously explored. In

addition to estimating own and cross export promotion elasticities, the magnitude of

cross effects on U.S. market share for grain commodities are simulated for several

policy scenarios involving alternative export promotion levels.

4.2 The Linear Approximation/Almost Ideal Demand System Model

(LA/AIDS)

The Rotterdam model, a major rival of the almost ideal demand system

(AIDS), is less preferred in the present context because it explains the level of demand

as opposed to market share. Given the FAS programs are intended to enhance the

U.S.’s competitive position in the global market place, having market share rather than

quantity as the dependent variable has certain advantages in exposition

In this study, we focus on U.S. grain export demand to the international market

rather than to specific countries. We select the major grain exporting country

combination instead of the aggregation of the rest of the world to be representative of

non-U.S. grain exporting competitors. Thus, U.S. grain market share in this study is

not as small of a proportion relative to non-U.S. grain market shares compared to

using all U.S. competitors.

The almost ideal demand system (AIDS) introduced by Deaton and

Muellbauer (1980) has become a popular method for modeling demand systems. The

44

AIDS model is consistent with homogeneity, adding-up, and symmetry restrictions

from consumer demand theory. Theoretically, the AIDS specification is derived

explicitly from a consumer cost minimization problem. It provides an arbitrary first

order approximation to any demand system and satisfies the axioms of choice (Deaton

and Muellbauer (1980)).

A popular alternative to this model is the linear approximation AIDS

(LA/AIDS). While some authors have pointed out some undesirable properties of

LA/AIDS such as biased parameter estimates of the Stone price index (e.g., Green and

Alston), Hanh concludes that LA/AIDS is a reasonable approximation for a demand

system. Its popularity is due to the simplicity of its estimation, and feasibility of

testing restrictions on homogeneity and symmetry (Hayes, Wahl, and Williams).

Satyanarayana, Wilson and Demcey argue that even though LA/AIDS model results in

biased parameter estimates for the Stone price index, the bias is less severe for demand

systems using aggregate data compared with micro-level data. This study uses

aggregate data. The following presents an overview of the dynamic LA/AIDS model

used in this study.

The demand system consists of three major grain commodities: rice, wheat,

and sorghum, and two regions: U.S. and non-U.S. The ith equation in a dynamic AIDS

model can be specified as

(4.1) wit = µi + γ ijj=1

6

∑ ln p jt + βi ln(M t /P) +ψ iwit−1 + εit ,

where M t = pit

i=1

6

∑ qit is the nominal total expenditure on the group of goods being

analyzed which includes U.S. and non-U.S. rice, wheat, and sorghum exports in year t.

Mt is assumed an exogenous variable because in the first stage, consumers first decide

45

how much to spend on agricultural imports such as horticultures, and grains. In the

second stage, fixed expenditure on grains (Mt) was allocated across supply sources and

products including U.S. and non-U.S. rice, wheat and sorghum. The interpretation of

the estimated export demand elasticities for grains is “conditional” (on the total U.S.

and non-U.S. grain expenditures). Therefore, the grain system we estimated is

conditional on U.S. expenditures on grains as a group with the implicit assumption

that grains are a weakly separable group11. P is the price index for the group; pjt is the

price of the jth good within the group in year t; wit is the conditional budget share of

total expenditure allocated to the ith good in year t (wit = pit qit / Mt ). Note that the

model is dynamic since it includes a lagged dependent variable for market share, and

thus both short- and long-run elasticities can be derived. The price index in the AIDS

model is defined as

(4.2) lnP =α0 + α j

j=1

6

∑ ln p j +1/2 γ iji=1

6

∑j=1

6

∑ ln pi ln p j

Using the price index in equation (4.2) often raises empirical difficulties,

especially when aggregate annual time –series data are used, and it is common to use

Stone’s (geometric) price index (Ps) instead of P, Green and Alston (1990).

(4.3) lnP s = wit

i=1

6

∑ ln pit

The LA/AIDS model uses Stone’s price index; Blanciforti and Green (1983a).

Green and Alston (1990) suggest that if prices are highly collinear, P may be

approximately proportional to Ps, i.e., P ≅ ζPs. If P is exactly (linearly) proportional to

11 Weak separability is the necessary and sufficient conditions for the second stage of two-stage budgeting (Deaton and Muellbauer (1980), pp. 122-124)

46

Ps, the LA/AIDS model can be used to estimate the parameters of the AIDS model and

can be written as

(4.4) wit =α i + γ ijj=1

6

∑ ln p jt + βi[lnM t − lnP s]+ψ iwit−1 + εit ,

where αi = (µi - βi lnζ). The LA/AIDS model is a semi-log functional form. The

expenditure variable in the model is deflated by world price index (WPI), but all the

prices are in nominal form. Export promotions of rice, wheat, and sorghum, and other

demand shifters i.e. trend can be incorporated by modifying the intercept as

(4.5) titititiii TGPROMWPROMRPROM τφδθαα ++++= , ,

where α’i = (µi - βilnζ) ; ttt GPROMandWPROMRPROM , are annual rice, wheat

and grain export promotion expenditures in year t, respectively. The export promotion

expenditures were deflated by world price index and multiplied by special drawing

rights (SDR) to reflect the impact of changes in exchange rates on the purchasing

power of U.S. promotion; tT is a trend term to capture steady movements of

unmodelled variables.

We posited the following LA/AIDS model to fit the time-series data by

substituting equation 4.3 and 4.5 into equation 4.4. Thus it generates the following

LA/AIDS model:

(4.6) wit =αi + γijj=1

6

∑ lnp jt +βi[lnMt − wit

i=1

6

∑ lnpit]+θiRPROMt +δiWPROMt

+φiGPROM t + τ iTt +ψiwit−1 + εit ,

47

for i = 1, 2, 3, 4, 5, and 6, which correspond to the three U.S. and three non-U.S.

commodities and 1−itw is a lagged ith market share to calculate the long-run impact.

Adding up, homogeneity, and Slutsky symmetry can be imposed on this

system as follows:

(4.7) α i

i=1

6

∑ =1; γ iji=1

6

∑ = 0; and βi

i=1

6

∑ = 0;

(4.8) γ ijj=1

6

∑ = 0, for i =1,2,…,n;

(4.9) jiij γγ = for all i ≠ j

Demand theory employs three sets of restriction on consumer demand: adding

up in equation 4.7, homogeneity (of degree zero in prices and expenditure) in equation

4.8, and Slutsky symmetry in equation 4.9. These three conditions are imposed in the

empirical model estimated here.

4.3 Data

The models were estimated using annual time-series data from 1975 to 2005

(see Table 4.1 for average values for some of this data). The price and export value

data were obtained from the Food and Agriculture Organization (FAO). The U.S. has

exported many grain varieties; however, wheat and sorghum are two of the largest in

terms U.S. export grain markets. Barley, for example, is comparatively smaller in

terms of U.S. market share accounting for only 2% of world barley exports in 2005.

Corn markets are by far too big to include in the system since the United States is the

largest corn exporter in the world and most of it is used as animal feed or industrial

use rather than human consumption as rice, wheat and sorghum is.

48

For non-U.S. rice, wheat, and sorghum, we chose the top exporters in term of

export values to be representative of U.S. competitors. For milled rice, the sum of

Thailand, Vietnam, China, Pakistan, and India export values were used, which

accounts for 74% of the rest of the world export values. For non-U.S. wheat market

share, the EU, Canada and Australia are aggregated, which accounts for 74% of the

rest of the world export values. Australia and Argentina account for 67 % of the world

export values for sorghum, and these two countries were combined to represent non-

U.S. sorghum exports.

Table 4.1: An average export volumes and values of major exporters, rest of the world and world, 1975-2005.

Rice Wheat Sorghum

Volume

(ton) Value

($1,000) Volume

(ton) Value

($1,000) Volume

(ton) Value

($1,000)

U.S.A 1,794,207.81 658,346.85 32,587,388.73 4,790,756.82 6,054,612.67 688,211.65

Thailand 4,369,689.77 1,225,870.17 - - - -

Vietnam 1,518,895.13 339,627.87 - - - -

China 1,320,463.27 293,139.77 - - - -

Pakistan 1,346,693.68 412,179.46 - - - -

India 1,440,506.39 505,153.61 - - - -

EU - - 27,629,325.28 4,220,559.05 - -

Canada - - 17,291,997.99 2,747,669.80 - -

Australia - - 12,511,210.33 1,857,572.52 446,684.06 52,221.78 Argentina

-

-

-

-

1,894,664.33

181,316.84

Non-U.S. 1 9,996,248.24 2,775,970.88 57,432,533.59 8,825,801.37 2,341,348.38 233,538.63

ROW 2 12,655,889.45 3,772,673.43 77,704,976.91 11,950,754.89 3,155,308.38 350,530.76

World 14,450,097.26 4,431,020.27 110,292,365.64 16,741,511.71 9,209,921.05 1,038,742.4

% *

1/2

0.79 0.74 0.74 0.74 0.74 0.67

* the percentage of Non-U.S. exporting to the world exporting 1 the summation of the combination of Non-U.S. exporters.

Source : FAO Statistics

49

Annual export promotion expenditures for rice, wheat and sorghum including

both government and private contributions were obtained from FAS. U.S. Wheat

Associates, a private sector firm, have contributed funds to FMD since 1975 and MAP

since 1986 to expand U.S. wheat export markets. U.S.A. Rice Federation/U.S. Rice

Producers and U.S. Grains Council joined the programs and started contributing funds

to FMD in 1984 and MAP in 1986. U.S. Grains Council mainly develops export

markets for U.S. barley, corn, and sorghum. We use these expenditures as a proxy for

U.S. sorghum export promotion even though it also includes expenditures for corn and

barley (hereafter, we refer to this as sorghum export promotion). Six jointly estimated

equations in a demand system including U.S. and non-U.S. rice, wheat and sorghum,

are estimated simultaneously from 1975 through 2005. Special Drawing Rights

(SDRs) were obtained from International Monetary Fund (IMF). Definition of

variables and summary statistics for the data are reported in Table 4.2.

50

Table 4.2: Variable definitions and summary statistics, 1975-2005 Variable Definition Mean Minimum Maximum s.d

w1 Market share for U.S. rice exports 0.0375 0.0242 0.0623 0.009

w2 Market share for non-U.S. rice exports 0.1491 0.0797 0.2833 0.059

w3 Market share for U.S. wheat exports 0.2692 0.1896 0.3907 0.063

w4 Market share for non-U.S. wheat exports 0.4908 0.3763 0.5890 0.055

w5 Market share for U.S. sorghum exports 0.0390 0.0237 0.0581 0.010

w6 Market share for non-U.S. sorghum exports 0.0142 0.0013 0.0410 0.013

p1 Nominal export price for U.S. rice, $/ton 367.94 241.50 529.15 59.14

p2 Nominal export price for non-U.S. rice, $/ton1 282.31 211.11 416.04 46.31

p3 Nominal export price for U.S. wheat, $/ton 146.03 99.64 202.74 23.19

p4 Nominal export price for non-U.S. wheat, $/ton2 155.60 121.63 200.16 21.84

p5 Nominal export price for U.S. sorghum, $/ton 113.04 81.03 154.51 15.48

p6 Nominal export price for non-U.S. sorghum, $/ton3 98.92 70.27 167.42 19.57

q1 Volume of U.S. rice exports, ton 1,794,207.81 1,359,597.00 2,399,030.00 278,803.44

q2 Volume of non-U.S. rice exports, ton4 9,947,251.63 3,419,045.00 19,968,028.59 5,094,302.86

q3 Volume of U.S. wheat exports, ton 32,587,388.73 25,104,457.83 45,106,672.00 5,886,270.23

q4 Volume of non-U.S. wheat exports, ton5 57,432,533.59 31,132,997.18 77,732,717.81 12,944,444.60

q5 Volume of U.S. sorghum exports, ton 688,211.64 389,411.00 1,173,796.00 173,935.47

q6 Volume of non-U.S. sorghum exports, ton6 233,538.62 27,557.00 701,496.00 207,440.66

RPROM* Nominal Export promotion expenditures for U.S. rice, $ 8,167,991.00 2,268,496.00 14,632,490.00 3,443,029.60

WPROM* Nominal Export promotion expenditures for U.S. wheat, $ 13,781,393.00 6,151,000.00 20,811,097.00 4,776,134.70

GPROM* Nominal Export promotion expenditures for U.S. sorghum, $ 12,773,844.00 1,399,959.00 17,970,408.00 4,480,967.40

SDR Special Drawing Right (SDR) 1.36 0.98 1.55 0.12

WPI World Price Index (WPI) 94.86 61.66 122.63 16.38

1,4 Average export prices and volumes of rice exported from Thailand, Vietnam, China, Pakistan and India during 1975-2005.

2,5 Average export prices and volumes of wheat exported from EU, Canada, and Australia during 1975-2005.

3,6 Average export prices and volumes of sorghum exported from Australia and Argentina during 1975-2005.

51

4.4 Econometric Results

The model was estimated using the seemingly unrelated regression (SUR)

method. Given that the U.S. such a large exporter in the grain market, the endogeneity

of nominal price was examined using Hausman test. The results indicated no

endogeneity occurred12

. The grain demand system includes U.S. and non-U.S. rice,

wheat, and sorghum. The model suggests the existence of first-order autocorrelation

(AR(1)). The Bresuch-Godfrey’s serial autocorrelation test identified statistically

significant first-order autocorrelation in non-U.S. rice, U.S. wheat, non-U.S. wheat

equations. When estimated without AR processes, the Durbin-h values of U.S. and

non-U.S. rice and wheat were -5.32, 4.73, -6.54 and 2.6, indicating an autocorrelation.

When an AR(1) term was included in the regression, its coefficient was -0.41

statistically significant at the 1% level (p-value is 0.012). Therefore, we estimated the

LA/AIDS model with an AR(1) process:

(4.10) εi,t = ρi1εit−1 +υ i,t ,

where 1iρ is the first-order autoregressive parameters and υ it is a white-noise

disturbance. None of six equations generated statistically significant heteroscedasticity

using White’s test 13

.

12 Wu-Hauman test for six individual equations. U.S.wheat and non-U.S. wheat price were tested for endogeneity because the suspicion of big market shares in the system. Each U.S wheat and non U.S wheat price were regressed on all exogenous variables including extra variables i.e. world GDP, SDR, and WPI to be as an instrumental variable. The coefficient of residual calculated previously from U.S. wheat price equation was not statistically significant in all 6 equations with p-value 0.55, 0.5, 0.35, 0.47, 0.23, and 0.52 respectively indicating no endogeneity. For non-U.S. wheat price, the p-values were 0.65, 0.25, 0.59, 0.29, 0.23, and 0.63 respectively. 13 The White’s test showed that there is no heteroskedasticity occurred in six equations. P-values of six equations were 0.44, 0.87, 0.43, 0.206, 0.39 and 0.46 respectively. Thus, all equations failed to reject the null hypothesis that there is no heteroskedasticity.

52

Since wit

i=1

6

∑ =1, only five equations are necessary to estimate the system.

The AR(1) LA/AIDS model was estimated imposing price homogeneity (equation 4.7

and 4.8) and symmetry (equation 4.9). The parameter estimates are reported in Table

4.3, Table 4.4 and Table 4.5.

The adjusted R2 of U.S. and non-U.S. rice, wheat and sorghum equations is

0.69, 0.81, 0.61, 0.42, 0.43 and 0.81, respectively. Five of the six own-price

coefficients and five of the fifteen cross-price coefficients are statistically significant

at the 5% or 10% level. Eleven of the eighteen export promotion coefficients are

significant at the 5% or 10% level. Own export promotion coefficients for rice, wheat

and sorghum were all positive and statistically significant in terms of increasing own

U.S. market share, however, cross export promotion coefficient signs are varied.

The trend term was found to be statistically significant for most grains except

non-U.S. wheat and non-U.S. sorghum. U.S. rice, wheat and sorghum have a negative

trend, while non-U.S. rice has a positive trend. Expenditure coefficients are

statistically significant at the 5% or 10% level on U.S. and non-U.S rice, non-U.S.

wheat and non-U.S. sorghum.

53

Table 4.3: Price coefficients of LA/AIDS for U.S. grains, 1975-2005.

Equations Market Share

Price Coefficients

γi1 γi2 γi3 γi4 γi5 γi6

U.S.Rice 0.0487 0.039***

(3.16)

Non-U.S.Rice 0.0978 -0.018 (-1.52)

0.065* (1.96)

U.S. Wheat 0.3251 0.053**

(2.38)

-0.072* (-1.74)

0.199* (1.82)

Non-U.S.Wheat 0.4614 -0.052**

(-2.39)

0.016 (0.37)

-0.170* (-1.72)

0.201* (1.74)

U.S. Sorghum 0.0533 -0.013 (-1.09)

0.003 (-0.19)

-0.023 (-0.889)

0.018 (0.690)

0.031* (1.756)

Non-U.S.Sorghum 0.0136 0.013 (0.945)

-0.018 (-1.44)

-0.038* (-1.68)

0.001 (0.07)

0.014 (0.791)

0.014 (1.44)

Table 4.4: Promotion coefficients of LA/AIDS for U.S. grains, 1975-2005.

Equations

Promotion Coefficients

U.S. Rice Promotion

θi

U.S. Wheat Promotion

δi

U.S. Sorghum Promotion

φi

U.S.Rice 0.009***

(2.66) 0.0067 (0.74)

-0.0085*** (-2.67)

Non-U.S.Rice -0.017* (-1.76)

-0.019 * (-1.76)

0.015 (1.42)

U.S. Wheat 0.017 (1.11)

0.093* (1.69)

-0.011 (-0.66)

Non-U.S.Wheat 0.024* (1.78)

-0.027 (-0.51)

-0.026* (-1.69)

U.S. Sorghum -0.006**

(-2.06) 0.004 (0.37)

0.008 ** (2.11)

Non-U.S.Sorghum -0.004* (-1.687)

-0.015* (-2.14)

0.0005 (0.18)

54

Table 4.5: Intercept, expenditure, trend, Adjusted R2 and Durbin – h value of LA/AIDS for U.S. grains, 1975-2005.

Based on the parameter estimates in Table 4.3, Table 4.4 and Table 4.5,

conditional own-price, cross price, expenditure and export promotion elasticities for

the LA/AIDS model were calculated from formulas derived in the literature as follows

15:

(4.11) iiiii wMq /1ln/ln βη +=∂∂=

(4.12) E ii = ∂lnqi /∂ln pi = −1+ γ ii /wi + wi

(4.13) E ij = ∂lnqi /∂ln p j = γ ij /wi + w j

(4.14) Ri = ∂lnqi /∂lnRPROMt = ∂lnwi /∂lnRPROMt = θi /wi

(4.15) W i = ∂lnqi /∂lnWPROM t = ∂lnwi /∂lnWPROM t = δi /wi

(4.16) Si = ∂lnqi /∂lnGPROMt = ∂lnwi /∂lnGPROMt = φi /wi ,

14 When the regression includes lagged dependent variables the Durbin-Watson d-statistic is not valid as a test for autocorrelated residuals. The d-statistic tends to be biased towards 2. In this case, Durbin's h test is more proper. For a test of the null hypothesis of no autocorrelation against the 2-sided alternative of autocorrelated errors, at a 5% level, the decision rule is if -1.96 < h < 1.96 do not reject the null hypothesis.

15 The elasticities are called “conditional” because they are computed under the hypothesis that group expenditure is exogenous. The formulas in equations (4.11) and (4.12) refer to compensated or Slutsky elasticities. Slutsky elasticities are used in preference to the Cournot or uncompensated elasticities because the formulas for the latter are more complicated, and the income effects of price changes are apt to be negligible owing to the fact grain products constitute a small portion of world income.

Equations Intercept Expenditure Trend Adj. R2 Durbin - h14

αi βi τ i

U.S.Rice 0.51* (1.92)

-0.04*** (-2.74)

-0.002** (-3.55)

0.69 -0.775

Non-U.S.Rice 1.47 (1.48)

-0.12** (-2.63)

0.004* (1.89)

0.81 1.63

U.S. Wheat -1.26 (-0.79)

0.02 (0.27)

-0.005** (-2.08)

0.61 -1.62

Non-U.S.Wheat -0.29

(-0.19) 0.099* (1.67)

0.002 (1.06)

0.42 0.77

U.S. Sorghum -0.09

(-0.27) 0.005 (0.32)

-0.0009** (-1.92)

0.43 -0.02

Non-U.S.Sorghum 0.04 (0.19)

0.016* (1.67)

0.0002 (-0.57)

0.81 0.40

55

where i =1, 2, 3, 4, 5, and 6 indicating U.S rice, non-U.S. rice, U.S. wheat, non-U.S.

wheat, U.S. sorghum, and non-U.S. sorghum respectively; iη is expenditure elasticity;

iiE is compensated or Slutsky own-price elasticity; ijE is compensated cross-price

elasticity;iR is rice export promotion elasticity;

iW is wheat export promotion

elasticity; iS is sorghum export promotion elasticity. The estimated short-run

compensated price elasticities are given in Table 4.6, and estimated short-run

promotion and expenditure elasticities are given in Table 4.7. The long-run elasticities

of compensated price and promotion and expenditure are listed in Table 4.8 and Table

4.9 respectively. We focus here on the own and cross promotion elasticities, which are

in Table 4.7 and Table 4.8 in the discussion that follows.

56

Table 4.6: Estimated short-run compensated price elasticities of U.S. and non-U.S. rice, wheat and sorghum, 1975 -2005

Table 4.7: Short-run compensated promotion and expenditure elasticities of U.S. and non-U.S. rice, wheat and sorghum, 1975 -2005

Equations SR Price Elasticity

Ei1 Ei2 Ei3 Ei4 Ei5 Ei6

U.S. Rice -0.14***

Non-U.S.Rice -0.29 -0.24 *

U.S. Wheat 1.43 ** -0.41 * -0.062 *

Non-U.S.Wheat -0.61 ** 0.63 -0.063 * -0.102 *

U.S. Sorghum -0.22 0.081 -0.017 0.094 -0.36 *

Non-U.S.Sorghum 0.28 -0.174 -0.103 * 0.016 0.275 0.035

Equations

SR Promotion Elasticity Expenditure Elasticity

iη U.S. Rice Ri

U.S.Wheat Wi

U.S.Sorghum Si

U.S. Rice 0.186*** 0.138 -0.174*** 0.197***

Non-U.S.Rice -0.178 * -0.202 ** 0.154 -0.259 **

U.S. Wheat 0.052 0.287 * -0.033 1.065

Non-U.S.Wheat 0.052 * -0.059 -0.057 * 1.214 *

U.S. Sorghum -0.115 ** 0.082 0.148 ** 1.105

Non-U.S.Sorghum -0.298 * -1.12 ** 0.04 2.200*

57

Table 4.8: Long- run compensated price, export promotion and expenditure elasticities of U.S. and non-U.S. rice, wheat and sorghum, 1975 -2005.

Table 4.9: Long- run promotion and expenditure elasticities of U.S. and non-U.S. rice, wheat and sorghum, 1975 -2005.

16 The long-run coefficient (λ i) of each grain equation was calculated by one subtracted by the

coefficient of each grain lagged market share variable (λ i = 1- ψi) which, for U.S. and non U.S. grain, were equal to 0.904, 0.46, 0.47, 0.62, 0.55 and 0.72 respectively. Long-run elasticity of price and promotion was obtained by dividing short run elasticity (Table 4.6 and 4.7) by the long-run coefficient

(λ i,).

Equations

LR Price Elasticity

Lagged

Market share ψi 16

Ei1 Ei2 Ei3 Ei4 Ei5 Ei6

U.S.Rice 0.096 -0.15 ***

Non-U.S.Rice 0.54*** -0.63 -0.51*

U.S. Wheat 0.53 *** 3.06 ** -0.87* -0.13 **

Non-U.S.Wheat 0.38*** -1.00** -0.103 -0.103 * -0.16 *

U.S. Sorghum 0.45*** -0.41 0.15 -0.03 0.17 -0.65 *

Non-U.S.Sorghum 0.28* 0.39 -0.24 -0.14 * -0.02 0.38 0.048

Equations

LR Promotion Elasticity LRExpenditure Elasticity

iη U.S. Rice Ri

U.S.Wheat Wi

U.S.Grains Si

U.S. Rice 0.205*** 0.153 -0.192*** 0.218***

Non-U.S.Rice -0.389 * -0.442 0.337 -0.567 **

U.S. Wheat 0.111 0.616 * -0.071 2.286

Non-U.S.Wheat 0.085 * -0.097 -0.092 * 1.986*

U.S. Sorghum -0.209 ** 0.149 0.269 ** 2.000

Non-U.S.Sorghum -0.414 * -1.558 ** 0.055 3.050*

58

In Table 4.7 (short-run) and Table 4.9 (long-run), the own effects of U.S.

export promotion for the three commodities are all positive and statistically

significant. U.S. wheat promotion has the largest estimated direct effect, with a short-

run elasticity of 0.287 and a long-run elasticity of 0.616. U.S. rice promotion has

short- and long-run elasticities of 0.186 and 0.205, and U.S. sorghum promotion’s

impact on U.S. sorghum market share has a short-run elasticity of 0.148 and a long-

run elasticity of 0.269 (note that since this promotion variable also includes corn and

barley expenditures it may be biased downwards). Therefore, it appears that U.S.

export promotion has been effective in increasing U.S. market share of wheat, rice,

and sorghum. These own effects are consistent with the majority of previous studies

that find a significant demand effect of export promotion (see Alston et al.).

Policy makers are also interested in whether there are “halo effects” of U.S.

export promotion. Our results indicate that there is no halo effects of U.S. export

promotion on other U.S. grains. U.S. rice export promotion has a positive impact, but

not statistically significant on U.S. wheat market share, with a short-run cross

elasticity of 0.052 (0.111, long-run) but a negative and statistically significant impact

on U.S. sorghum market share with a cross elasticity of -0.115 (-0.209, long-run).

These two short-run cross elasticities sum to -0.063 (-0.098 long-run), but are not

statistically significant based on a Wald test (p-value is 0.7). U.S. wheat export

promotion has no significant impact on either U.S. rice (0.138, short-run; 0.153, long-

run) or U.S. sorghum (0.082, short-run; 0.149, long-run) market share. Again, a Wald

test (p-value is 0.47) indicates no halo effect from U.S. wheat promotion on other U.S.

grains. U.S. sorghum promotion has a negative and statistically significant impact on

U.S. rice (-0.174, short-run; -0.192, long-run), but not statistically significant on U.S.

wheat (-0.033, short-run; -0.071, long-run) market share. However, we fail to reject

the null hypothesis that the two cross elasticities sum to zero (p-value in Wald test is

59

0.27). In terms of statistically significant cross elasticities, we found two negative

impacts, and no positive impacts. Hence, our results are the opposite of Richards and

Patterson’s findings of a halo effect for horticultural commodities.

Another important policy issue in export promotion is whether other countries

free ride off of U.S. promotion, i.e., whether there is a halo effect of U.S. promotion

for other countries. Our results indicate that U.S. export promotion actually hurts

competing country exports. U.S. rice export promotion has a negative and statistically

significant impact on rice (-0.178, short-run; -0.389, long-run) and sorghum (-0.298,

short-run; -0.414 long-run) market share for competing countries, and a positive and

statistically significant impact on wheat (0.052, short-run; 0.085 long-run) market

share for competing countries. Collectively, these three estimated cross elasticities

sum to -0.424 (-0.718, long-run), indicating a negative halo effect of U.S. rice

promotion on non-U.S. grain demand. Based on a Wald test, we reject the null

hypotheses of no anti-halo effect at the 10 percent significance level (p-value is 0.08).

U.S. wheat export promotion has a negative and statistically significant effect on rice

(-0.202, short-run; -0.442, long-run) and sorghum (-1.12, short-run; -1.558, long-run)

market share of non-U.S. countries, and a negative, but not statistically significant

impact on wheat (-0.059, short-run; -0.097) market share of non-U.S. countries.

Collectively, these three cross elasticities sum to -1.381 (-2.097 long-run), and, based

on a Wald test, are statistically significant (p-value is 0.07). Consequently, there is an

anti-halo effect of U.S. wheat promotion on the demand for grains of other countries.

U.S. sorghum export promotion has a negative and statistically significant impact on

wheat (-0.057, short-run; -0.092, long-run) market share of non-U.S. countries, and a

positive, but not statistically significant impact on rice (0.154, short-run; 0.337, long-

run) and sorghum (0.04, short-run; 0.055, long-run) market share of non-U.S.

countries. Based on a Wald test, we could not reject the null hypothesis of no anti-

60

halo effect at the 10 percent significance level (p-value is 0.58). In terms of

statistically significant individual cross elasticities, there are five significant negative

elasticities and only one significant positive elasticity. This provides empirical

evidence of an anti-halo effect of U.S. grain export promotion on competing country

grain exports.

In terms of the econometric results, there are three general conclusions. First,

U.S. grain export promotion has a positive and significant direct impact on U.S. grain

exports for rice, wheat, and sorghum. Second, while the direct effects are significant,

there are no halo effects of U.S. grain export promotion on other U.S. grains. Finally,

U.S. grain export promotion has an anti-halo effect on competing country grain

exports. Hence, there is no international free riding off of U.S. grain export

promotion.

4.5 Simulations

In order to examine the benefits and costs of grain export promotion, several

simulations were performed. For each grain, the LA/AIDS model was used to

simulate two scenarios: (1) a baseline scenario with U.S. grain export promotion

expenditures at historical levels – export promotion programs are in effect; and (2) a

counterfactual scenario with U.S. grain export promotion reduced by 50 percent. In

the first scenario, all grain export demand determinants are set equal to their historic

levels and simulated over the period 1990-2005. The second scenario is identical to

the first, except U.S. grain export promotion expenditures are set to 50 percent of their

historical amounts. These two scenarios are run separately for rice, sorghum, and

wheat. The difference between the two scenarios for each grain gives the total impact

of the U.S. grain export promotion programs on U.S. grain market share.

61

These results are then translated into a benefit-cost ratio using the following

procedures. First, U.S. market share (wit) for each grain and each scenario is

transformed into gross value of U.S. exports (Vit) by multiplying market share by total

U.S. and non-U.S. grain expenditures (Mt). The gross value is then translated into a

net value (vit) by multiplying each grain by a net margin factor equal to ititit pcp /)( − ,

for i = rice, wheat, and sorghum for t = 1990-2005, and where: pit is the unit price and

cit is the unit cost of grain i. Next, a “direct” benefit-cost ratio (DBCR) is calculated as

follows:

(4.17) )/()( ''

ititititit EPCOSTEPCOSTvvDBCR −−= ,

where vit is the U.S. export value of grain i for scenario 1 (i=rice, wheat, and

sorghum); vit’ is the U.S. export value of grain i for scenario 2; EPCOSTit is export

promotion expenditures for grain i under scenario 1; and EPCOSTit’ is export

promotion expenditures for grain i under scenario 2. DBCRit provides a measure of

the direct benefit-cost ratio for export promotion for grain i.

We also compute an alternative benefit-cost ratio that captures the direct and

the cross effects of promotion. This total BCR (TBCR) is calculated as follows:

(4.18)

TBCRit = ( v jt

j=1

3

∑ − v jt

' ) /( EPCOSTjt

j=1

3

∑ − EPCOSTjt

' ),

where i = rice, wheat, and sorghum.

The DBCR is a useful measure for producers in each specific industry.

However, for policy makers, the TBCR is probably a better measure since it gives the

over all impact of a specific export promotion on the entire three grain industries.

62

The results are displayed in Table 4.10. U.S. rice export promotion is

beneficial to both the rice industry, and to all three-grain sectors combined. On

average over the period 1990-2005, the DBCR is equal to 5.34, indicating that each

dollar invested in rice export promotion had resulted in $5.34 in net export revenue to

the U.S. rice sector. The average TBCR for U.S. rice export promotion for this period

of even higher at 5.46 indicating positive net revenue effects of U.S. rice promotion

for all three grains combined.

Table 4.10: An average direct and total benefit-cost ratios for U.S. rice, wheat, and sorghum export promotion (1990-2005):

U.S. wheat export promotion was also beneficial for both the wheat and

combined three grain industries, and had the highest benefit-cost ratios. The DBCR

for wheat over this time period was computed to be 26.10, i.e., each dollar resulted in

$26.10 in additional net export revenue to the U.S. wheat industry. The TBCR was

even higher at 30.10 indicating spill-over benefits to the three grain sectors combined.

U.S. sorghum promotion had mixed impacts. In terms of the DBCR, U.S.

sorghum export promotion was beneficial to U.S. sorghum producers. Each dollar

invested returned, on average, $5.18 in terms of net export revenue. However, U.S.

sorghum export promotion actually had a negative impact on combined net revenue

for the three grains. Each dollar invested in U.S. sorghum promotion resulted in a loss

of net export revenue of $0.72.

In summary, in terms of BCRs, the direct effects of U.S. rice, wheat, and

sorghum export promotion is benefiting grain producers in the United States. In terms

Grains Direct Benefit-Cost Ratio Total Benefit-Cost Ratio

Rice 5.34 5.46 Wheat 26.10 30.10 Sorghum 5.18 -0.72

63

of total effects, U.S. rice and wheat promotion is beneficial to rice, wheat, and

sorghum producers combined, but U.S. sorghum promotion is not benefiting all three

grains combined.

4.6 Summary and Conclusions

In this article, we investigated both the own and cross effects of U.S. export

promotion for U.S. and non-U.S. grains. The focus was on two previously neglected

aspects of the cross-effects, whether: (1) a halo effect for U.S. commodities exists,

and (2) U.S. grain export promotion has a halo effect on grain demand for other

countries? An LA/AIDS model for U.S. and non-U.S. rice, wheat, and sorghum that

include U.S. export promotion as one of the explanatory variables was used to

examine these issues.

U.S. grain export promotion was found to have a positive and significant direct

impact on U.S. grain exports for rice, wheat, and sorghum. Wheat promotion was

found to have the largest impact on U.S. market share, followed by rice, and sorghum

promotion. Regarding indirect effects on U.S. grain demand on rice, wheat, and

sorghum export promotion were not found to have a statistically significant positive,

i.e., halo effect. On the contrary, in terms of statistically significant cross elasticities,

we found two negative results, indicating anti-halo effect. Hence, our results are the

opposite of Richards and Patterson’s findings of a halo effect for horticultural

commodities. U.S. grain export promotion was found to have an anti-halo effect on

competing country grain exports. In terms of statistically significant individual cross

elasticities, there were five significant negative elasticities and only one significant

positive elasticity. This provides empirical evidence of an anti-halo effect of U.S.

grain export promotion on competing country grain exports. Hence, there is no

international free-riding off of U.S. grain export promotion.

64

In order to examine the benefits and costs of grain export promotion, several

simulations were performed. For each grain, the LA/AIDS model was used to

simulate two scenarios: (1) a baseline scenario with U.S. grain export promotion

expenditures at historical levels – export promotion programs are in effect; and (2) a

counterfactual scenario with U.S. grain export promotion reduced by 50 percent. The

difference between the two scenarios for each grain gives the total impact of the U.S.

grain export promotion programs. Two benefit-cost ratios were computed for each of

the three grains.

First, a direct benefit-cost ratio on average over period 1990-2005 was

computed, which provides a measure of the direct benefit-cost ratio for export

promotion for each grain. For all three grains, the direct BCR was significantly

higher than 1.0 indicating positive net benefits to producers in each of the three-grain

industries. Wheat had the largest direct BCR at 26.10, followed by rice (5.34) and

sorghum (5.18). Second, an alternative benefit-cost ratio that captures the direct and

the cross effects of promotion were computed. Wheat and rice had total BCRs that

were higher than the direct BCR indicating halo effects, while sorghum actually had a

negative BCR indicating anti-halo effect of U.S. grain export promotion.

65

REFERENCES

Alston, J.M. and Crespi, J.M and Kaiser, H.M. and Sexton, R.J. The Economics of

Commodity Promotion Programs: Summary and Synthesis. In: Kaiser, H.M.,

Alston, J.M., Crespi, J.M. and Sexton, R.J.(eds). The Economics of Commodity

Promotion Programs: Lessons from California. Peter Lang, New York (2005). pp.

387-415.

Asche, F. and Wessells, C.R. (1997). On Price Indices in the Almost Ideal Demand

System. American Journal Agricultural Economics (79). pp. 1182-1185.

Chang, Hui-Shung and Nguyen, C. Elasticity of Demand for Australian Cotton in Japan.

The Australian Journal of Agricultural and Resource Economics 46(1). pp. 99-113.

Deaton, A. and Muellbauer, J. (1980a). Economics and Consumer Behavior. Vail-Ballou

Press, New York.

Dwyer, M.J. (1995). Evaluating the Effectiveness of the Market Promotion Program on

U.S. High-Value Agricultural Exports. Final Regulatory Impact Analysis. FAS staff

paper. USDA, Washington, D.C. pp. 1-95.

Food and Agriculture Organization (FAO) of the United Nations. FAO statistics- Agriculture

(2007).

Green, R. and Alston, J.M. (1990). Elasticities in AIDS Models. American Journal of

Agricultural Economics(72). pp. 442-445.

Green, R. and Alston, J.M. (1991). Elasticities in AIDS Models : A Clarification and

Extension. American Journal of Agricultural Economics(73). pp. 874-875.

Greene, W.H. Econometric Analysis (5th edition). Pearson Education, Inc.,(2003). Time-

series models. pp. 608-662.

Hahn, W.F. (1994). Elasticities in AIDS models: Comment. American Journal of

Agricultural Economics(76). pp. 972-977.

66

Hayes, D.J. and Wahl, T.I. and Williams, G.W. (1990). Testing Restrictions on a Model of

Japanese Meat Demand. American Journal of Agricultural Economics72(3). pp.

556-566.

International Monetary Fund (IMF). International Financial Statistics.

Johnson, A.J. and Durham, C.A. and Wessells, C.R. (1998). Seasonality in Japanese

Household Demand for Meat and Seafood. Agribusiness14(4). pp. 337-351.

Kinnucan, H.W. and Zheng, Y. (2005). National Benefit-Cost Estimations for the Dairy,

Beef, Pork and Cotton Promotion Programs. In: Kaiser, H.M., Alston, J.M. ,

Crespi,J. and Sexton, R.J.(eds). The Economics of Commodity Promotion Programs:

Lessons from California. Peter Lang, New York. pp. 255-284.

Richards, T.J. and Patterson, P.M. (1998).Dynamic Complementarity in Export Promotion

: The Market Access Program in Fruits and Vegetables. Journal of Agricultural

and Resource Economics 23(2). pp. 319-337.

Richard, T.J. and Ispelen, P.V. and Kagan, A.(1997).A Two-Stage Analysis of the

Effectiveness of Promotion Programs for U.S. Apples. American Journal of

Agricultural Economics 79(3). pp. 825-837.

Satyanarayana, V. and Wilson, W.W. and Johnson, D.D. (1999). Import Demand for Malt

in Selected Countries : A Linear Approximation of AIDS. Canadian Journal of

Agricultural Economics (47). pp. 137-149.

The U.S. Grains Council.

USDA Foreign Agricultural Service (FAS).

Xiao, H. and Kinnucan, H.W. and Kaiser, H.M. (1999). Effects of Advertising on U.S. Non-

Alcoholic Beverage Demand: Evidence form a Rotterdam Model. Journal of

Agricultural and Resource Economics.

67

CHAPTER 5

SUMMARY AND CONCLUSIONS

The objectives of this thesis were to: (1) to empirically measure the

responsiveness of U.S. rice and other related grain exports in term of demand aspect

and market share aspect with respect to U.S. grain export promotions, and (2) to assess

the overall effectiveness of the U.S. grain export programs in the world market in term

of benefits relative to costs. The thesis was comprised of two essays: (1) Measuring

the Effectiveness of U.S. Rice Export Promotion Programs and (2) The Halo Effects of

U.S. Grain Export Promotion.

The first essay – “ Measuring the Effectiveness of U.S. Rice Export Promotion

Programs” measured the export promotion elasticity by estimating U.S. rice export

demand using an instrumental-variable regression approach and a single equation with

double logarithmic function given other demand determinants such as own price, price

of competing countries, income, and exchange rates and U.S. rice export promotion

expenditures. There were three main objectives of the essay. The first objective was to

econometrically investigate the responsiveness of U.S. rice export demand with

respect to U.S. rice export promotion. The second objective was to evaluate the

effectiveness of U.S. rice export promotion programs based on benefit-cost ratio

(BCR) outline. The third objective was to estimate the optimality of U.S. rice export

promotion expenditure levels by computing a marginal benefit-cost ratio (MBCR).

Over the period 1984–2005, the results support the hypothesis that U.S. rice

export promotion programs have had a positive impact on the U.S. rice export

demand. The export promotion elasticity was computed to be 0.143, which was

statistically significantly different from zero.

To address the question of net return of export promotion investment, the

simulated benefits of export promotion were compared against the costs of the

68

programs and the associated benefit-cost ratios then were computed. The resulting

average benefit-cost ratios of the U.S. rice export promotion expenditures were

computed based on a range of excess supply own price elasticities (from 1 to 10). The

average BCRs ranged from 30.05 for the most inelastic estimate, to 5.51 for the most

elastic. All of these values were well above 1.0. Thus, the main policy conclusion was

that the benefits of export promotion programs in terms of enhancing producer welfare

have been much greater than the costs of the programs.

To explore the optimality of U.S. rice export promotion, the marginal BCRs

were estimated between 9.61 and 1.76 given the own-price elasticity of excess supply

between 1 and 10. All results were larger than 1.0 indicating that the U.S. is under-

investing in U.S. rice export promotion from an optimality standpoint.

The second essay – “The Halo Effects of U.S. Grain Export Promotion”

focused on spill-in and spill-over effects of U.S. grain export promotion and the

characteristics of U.S. grain export promotion as a international public good over the

period 1975-2005. Linear approximation AIDS (LA/AIDS) with seemingly

uncorrelated regression (SUR) method was presented to estimate a demand system

consisting of three major grain commodities: rice, wheat, and sorghum, and two

regions: U.S. and non-U.S. The model is consistent with homogeneity, adding-up, and

symmetry restrictions from consumer demand theory. There were two main objectives

of the essay. The first objective was to examine both the own and cross effects of U.S.

grain export promotion for U.S. and non-U.S. grains. The second objective was to

evaluate the magnitude of cross effects on U.S. grain market shares by simulating

several policy scenarios involving alternative export promotion levels and then

computing direct and total benefit-cost ratios (DBCR, TBCR).

U.S. grain export promotion was found to have a positive and significant direct

effect on U.S. grain market shares in rice, wheat, and sorghum; however, indirect

69

effects on U.S. grain market shares were not found to have a statistically significant

positive, i.e. no halo effect. On the contrary, in terms of statistically significant cross

elasticities, we found two negative results, indicating an anti-halo effect.

In terms of international free riding, the results indicated that U.S. grain export

promotion has an anti-halo effect on competing country grain exports indicating five

significant negative elasticities and only one significant positive elasticity. Hence,

there is no international free-riding off of U.S. grain export promotion.

Benefits and costs of U.S. grain export promotion were computed with several

simulations. Baseline scenario with U.S. grain export promotion expenditures at

historical levels and a counterfactual scenario – 50% reduction of U.S. grain export

promotion expenditures were set. The difference gives the total impact of U.S. grain

export promotion programs. Direct benefit-cost ratios (DBCR) of three-grain

promotions were greater than one with wheat (26.10), rice (5.34) and sorghum (5.18).

Total benefit-cost ratios (TBCR) gave higher value than DBCR in rice and wheat

export promotion while sorghum export promotion had a negative total benefit-cost

ratio. In conclusion, U.S. rice and wheat export promotion is benefiting grain

producers, but U.S. sorghum export promotion is not.

In summary, both essays concluded that U.S. grain export promotion has a

positive and statistically significant direct effect on rice, wheat and sorghum. Indirect

effects which were estimated in the second essay indicated that there are no halo

effects of U.S. export promotion on other U.S. grains. U.S. grain export promotion

also provides empirical evidence of anti-halo effects of U.S. grain export promotion

on competing country grain exports. In term of effectiveness of U.S. grain export

promotion, direct benefit-cost ratios (DBCR) were greater than 1.0 for rice, wheat and

sorghum export promotion; total benefit-cost ratios (TBCR) were positive in rice and

wheat export promotion, but negative in sorghum export promotion.